{"altLabels":["Ad Hoc Search","Adhoc Retrieval"],"definition":"A retrieval task in which a one-off query or topic is submitted against a fixed corpus or evaluation dataset to find topically relevant items.","derivedRelations":[{"note":"Known-item search targets a particular item already in the user's mind; ad hoc retrieval usually targets topical relevance for a new information need.","target":"known-item-search","type":"notEquivalentTo"},{"note":"Session search considers multiple interactions over time; ad hoc retrieval is often framed as a single query against a fixed collection.","target":"session-search","type":"notEquivalentTo"},{"note":"Ad hoc retrieval is a classic task formulation that is often evaluated in a single-shot way, but single-shot retrieval names the non-iterative interaction pattern rather than the collection/topic setup.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"note":"Tip-of-the-tongue search seeks a specific previously encountered item; ad hoc retrieval usually seeks topically relevant material for a new information need.","target":"tip-of-the-tongue-search","type":"notEquivalentTo"}],"distinctions":[{"reason":"Ad hoc retrieval is usually framed around retrieving documents or passages for a topic; entity search targets entities or entity-oriented information.","statement":"Ad Hoc Retrieval is not Entity Search.","target":"entity-search","targetLabel":"Entity Search","targetUri":"https://id.searchplex.net/entity-search/"},{"reason":"Ad hoc retrieval assumes a known collection; federated search searches across multiple sources or search services and must combine their results.","statement":"Ad Hoc Retrieval is not Federated Search.","target":"federated-search","targetLabel":"Federated Search","targetUri":"https://id.searchplex.net/federated-search/"},{"reason":"Information filtering continuously selects incoming items for stable interests or profiles; ad hoc retrieval evaluates one-off topics against a known collection.","statement":"Ad Hoc Retrieval is not Information Filtering.","target":"information-filtering","targetLabel":"Information Filtering","targetUri":"https://id.searchplex.net/information-filtering/"},{"reason":"Ad hoc retrieval can be evaluated as a single-shot batch task; interactive search emphasizes user-system interaction across a search session.","statement":"Ad Hoc Retrieval is not Interactive Search.","target":"interactive-search","targetLabel":"Interactive Search","targetUri":"https://id.searchplex.net/interactive-search/"},{"reason":"Routing applies persistent information needs to new incoming documents; ad hoc retrieval applies new one-off information needs to a known collection.","statement":"Ad Hoc Retrieval is not Routing.","target":"routing","targetLabel":"Routing","targetUri":"https://id.searchplex.net/routing/"}],"id":"ad-hoc-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC 1999 Adhoc Dataset","type":"officialHomepage","url":"https://catalog.data.gov/dataset/trec-1999-adhoc-dataset"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Ad Hoc Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Ad hoc retrieval is usually framed around retrieving documents or passages for a topic; entity search targets entities or entity-oriented information.","target":"entity-search","type":"notEquivalentTo"},{"note":"Ad hoc retrieval assumes a known collection; federated search searches across multiple sources or search services and must combine their results.","target":"federated-search","type":"notEquivalentTo"},{"note":"Information filtering continuously selects incoming items for stable interests or profiles; ad hoc retrieval evaluates one-off topics against a known collection.","target":"information-filtering","type":"notEquivalentTo"},{"note":"Ad hoc retrieval can be evaluated as a single-shot batch task; interactive search emphasizes user-system interaction across a search session.","target":"interactive-search","type":"notEquivalentTo"},{"note":"Routing applies persistent information needs to new incoming documents; ad hoc retrieval applies new one-off information needs to a known collection.","target":"routing","type":"notEquivalentTo"},{"target":"document-retrieval","type":"related"},{"target":"ranking","type":"related"},{"target":"routing","type":"related"},{"target":"single-shot-retrieval","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/ad-hoc-retrieval/","json":"https://id.searchplex.net/ad-hoc-retrieval.json","jsonld":"https://id.searchplex.net/ad-hoc-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Ad hoc retrieval is the classic TREC-style retrieval task: the searchable corpus or test-collection dataset is fixed for evaluation, while each query or topic represents a new information need. It is usually framed around topical document or passage relevance rather than ongoing filtering, source selection, entity lookup, or interactive exploration.","status":"published","uri":"https://id.searchplex.net/ad-hoc-retrieval/"}
{"altLabels":["Agentic Retrieval-Augmented Generation"],"definition":"A retrieval-augmented generation pattern in which an agent plans, issues, evaluates, and iterates retrieval or tool-use actions before producing a generated answer or result.","distinctions":[{"reason":"Agentic retrieval names the retrieval loop; Agentic RAG is the broader generation pattern that uses such a loop to produce an answer or result.","statement":"Agentic RAG is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Agentic RAG is a retrieval-augmented generation pattern; deep research emphasizes multi-step research, browsing, evidence synthesis, and often report-style output.","statement":"Agentic RAG is not Deep Research.","target":"deep-research","targetLabel":"Deep Research","targetUri":"https://id.searchplex.net/deep-research/"},{"reason":"Agentic RAG is a form of RAG with agentic control over retrieval or tool-use steps; RAG in general may use a fixed retrieve-then-generate pipeline.","statement":"Agentic RAG is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"},{"reason":"Retrieval planning is one capability within an agentic RAG system; it is not the full RAG pattern.","statement":"Agentic RAG is not Retrieval Planning.","target":"retrieval-planning","targetLabel":"Retrieval Planning","targetUri":"https://id.searchplex.net/retrieval-planning/"}],"id":"agentic-rag","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Agentic Retrieval-Augmented Generation","type":"canonicalSurvey","url":"https://arxiv.org/abs/2501.09136"},{"label":"From Web Search towards Agentic Deep Research","type":"canonicalSurvey","url":"https://arxiv.org/abs/2506.18959"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Agentic RAG","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Agentic retrieval names the retrieval loop; Agentic RAG is the broader generation pattern that uses such a loop to produce an answer or result.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Agentic RAG is a retrieval-augmented generation pattern; deep research emphasizes multi-step research, browsing, evidence synthesis, and often report-style output.","target":"deep-research","type":"notEquivalentTo"},{"note":"Agentic RAG is a form of RAG with agentic control over retrieval or tool-use steps; RAG in general may use a fixed retrieve-then-generate pipeline.","target":"rag","type":"notEquivalentTo"},{"note":"Retrieval planning is one capability within an agentic RAG system; it is not the full RAG pattern.","target":"retrieval-planning","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"deep-research","type":"related"},{"target":"grounding","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"query-routing","type":"related"},{"target":"rag","type":"related"},{"target":"retrieval-planning","type":"related"}],"representations":{"html":"https://id.searchplex.net/agentic-rag/","json":"https://id.searchplex.net/agentic-rag.json","jsonld":"https://id.searchplex.net/agentic-rag.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Agentic RAG combines RAG with agentic control over retrieval, query reformulation, tool use, evidence gathering, or stopping decisions. It is distinct from RAG in general, which may use a fixed retrieval pipeline, and from agentic retrieval, which names the retrieval loop rather than the full generation pattern.","status":"published","terms":[{"community":"industry","label":"Agentic RAG System","usage":"nearSynonym"},{"community":"industry","label":"Agentic RAG Workflow","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/agentic-rag/"}
{"altLabels":["Agentic Search"],"definition":"A retrieval pattern in which an AI agent plans, issues, evaluates, and iterates retrieval actions as part of a broader reasoning or task-solving loop.","derivedRelations":[{"target":"browsecomp-plus","type":"evaluatedBy"},{"target":"retrieval-planning","type":"hasPart"},{"note":"Conversational search centers on user-system dialogue; agentic retrieval centers on an agent controlling retrieval actions.","target":"conversational-search","type":"notEquivalentTo"},{"note":"Deep research is a broader agentic research and synthesis pattern; agentic retrieval is the retrieval loop that may support it.","target":"deep-research","type":"notEquivalentTo"},{"note":"Multi-hop retrieval is a retrieval pattern; agentic retrieval is the broader agent-controlled loop that may perform multi-hop retrieval.","target":"multi-hop-retrieval","type":"notEquivalentTo"},{"note":"Query routing chooses a path for a query; agentic retrieval describes an agent-controlled retrieval loop that may include routing decisions.","target":"query-routing","type":"notEquivalentTo"},{"note":"Retrieval planning is one capability within agentic retrieval; agentic retrieval is the broader agent-controlled retrieval pattern.","target":"retrieval-planning","type":"notEquivalentTo"},{"note":"A web browsing agent is an agent role or system; agentic retrieval is the retrieval behavior it may perform.","target":"web-browsing-agent","type":"notEquivalentTo"},{"target":"browsecomp","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"web-browsing-agent","type":"related"}],"distinctions":[{"reason":"Agentic retrieval names the agent-controlled retrieval loop; Agentic RAG is the broader generation pattern that uses such a loop.","statement":"Agentic Retrieval is not Agentic RAG.","target":"agentic-rag","targetLabel":"Agentic RAG","targetUri":"https://id.searchplex.net/agentic-rag/"},{"reason":"Agentic retrieval is agent-controlled; iterative retrieval is broader and may be user-driven, system-driven, feedback-driven, or agent-driven.","statement":"Agentic Retrieval is not Iterative Retrieval.","target":"iterative-retrieval","targetLabel":"Iterative Retrieval","targetUri":"https://id.searchplex.net/iterative-retrieval/"},{"reason":"Multi-stage ranking is a retrieval-system architecture; agentic retrieval is a task-loop pattern in which an agent may call one or more retrieval systems repeatedly.","statement":"Agentic Retrieval is not Multi-Stage Ranking.","target":"multi-stage-ranking","targetLabel":"Multi-Stage Ranking","targetUri":"https://id.searchplex.net/multi-stage-ranking/"},{"reason":"RAG is a generation pattern using retrieved information; agentic retrieval is an agent-controlled retrieval pattern that may be used within agentic RAG systems.","statement":"Agentic Retrieval is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"},{"reason":"Retrieval selects items for a query; agentic retrieval describes an agent-controlled loop of retrieval actions and result handling.","statement":"Agentic Retrieval is not Retrieval.","target":"retrieval","targetLabel":"Retrieval","targetUri":"https://id.searchplex.net/retrieval/"},{"reason":"Semantic search describes meaning-oriented retrieval; agentic retrieval describes who controls and iterates the retrieval process.","statement":"Agentic Retrieval is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Agentic retrieval iterates retrieval actions under agent control; single-shot retrieval uses one query or retrieval attempt without an agentic loop.","statement":"Agentic Retrieval is not Single-Shot Retrieval.","target":"single-shot-retrieval","targetLabel":"Single-Shot Retrieval","targetUri":"https://id.searchplex.net/single-shot-retrieval/"}],"id":"agentic-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"This is what agentic retrieval looks like","type":"officialHomepage","url":"https://hornet.dev/blog/this-is-what-agentic-retrieval-looks-like"},{"label":"Agentic Retrieval-Augmented Generation","type":"canonicalSurvey","url":"https://arxiv.org/abs/2501.09136"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Agentic Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Agentic retrieval names the agent-controlled retrieval loop; Agentic RAG is the broader generation pattern that uses such a loop.","target":"agentic-rag","type":"notEquivalentTo"},{"note":"Agentic retrieval is agent-controlled; iterative retrieval is broader and may be user-driven, system-driven, feedback-driven, or agent-driven.","target":"iterative-retrieval","type":"notEquivalentTo"},{"note":"Multi-stage ranking is a retrieval-system architecture; agentic retrieval is a task-loop pattern in which an agent may call one or more retrieval systems repeatedly.","target":"multi-stage-ranking","type":"notEquivalentTo"},{"note":"RAG is a generation pattern using retrieved information; agentic retrieval is an agent-controlled retrieval pattern that may be used within agentic RAG systems.","target":"rag","type":"notEquivalentTo"},{"note":"Retrieval selects items for a query; agentic retrieval describes an agent-controlled loop of retrieval actions and result handling.","target":"retrieval","type":"notEquivalentTo"},{"note":"Semantic search describes meaning-oriented retrieval; agentic retrieval describes who controls and iterates the retrieval process.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Agentic retrieval iterates retrieval actions under agent control; single-shot retrieval uses one query or retrieval attempt without an agentic loop.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"conversational-search","type":"related"},{"target":"deep-research","type":"related"},{"target":"interactive-search","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"multi-stage-ranking","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-routing","type":"related"},{"target":"rag","type":"related"},{"target":"semantic-search","type":"related"},{"target":"session-search","type":"related"},{"target":"single-shot-retrieval","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"}],"representations":{"html":"https://id.searchplex.net/agentic-retrieval/","json":"https://id.searchplex.net/agentic-retrieval.json","jsonld":"https://id.searchplex.net/agentic-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Agentic retrieval emphasizes multi-step and adaptive retrieval behavior by an agent, often involving query reformulation, tool use, filtering, verification, or repeated searches. It is not the same as retrieval generally, RAG, semantic search, or multi-stage ranking.","status":"published","terms":[{"community":"industry","label":"Agent Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/agentic-retrieval/"}
{"altLabels":["ANN Search","ANNS"],"definition":"A search method that returns near-neighbor results using an approximation strategy to trade some exactness for lower latency, memory use, or computational cost.","derivedRelations":[{"target":"reverse-image-search","type":"related"}],"distinctions":[{"reason":"Approximate nearest neighbor search is an efficiency method for neighbor lookup; dense retrieval is a retrieval approach using dense representations.","statement":"Approximate Nearest Neighbor Search is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"HNSW is one approximate nearest neighbor method; approximate nearest neighbor search is the broader method family.","statement":"Approximate Nearest Neighbor Search is not HNSW.","target":"hnsw","targetLabel":"HNSW","targetUri":"https://id.searchplex.net/hnsw/"},{"reason":"Approximate nearest neighbor search is one family of methods used in vector search; vector search is the broader practitioner term for retrieval over vector representations.","statement":"Approximate Nearest Neighbor Search is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"approximate-nearest-neighbor-search","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Approximate Nearest Neighbor Search in High Dimensions","type":"canonicalSurvey","url":"https://arxiv.org/abs/1806.09823"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Approximate Nearest Neighbor Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Approximate nearest neighbor search is an efficiency method for neighbor lookup; dense retrieval is a retrieval approach using dense representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"HNSW is one approximate nearest neighbor method; approximate nearest neighbor search is the broader method family.","target":"hnsw","type":"notEquivalentTo"},{"note":"Approximate nearest neighbor search is one family of methods used in vector search; vector search is the broader practitioner term for retrieval over vector representations.","target":"vector-search","type":"notEquivalentTo"},{"target":"dense-retrieval","type":"related"},{"target":"hnsw","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"vector-search","type":"related"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/approximate-nearest-neighbor-search/","json":"https://id.searchplex.net/approximate-nearest-neighbor-search.json","jsonld":"https://id.searchplex.net/approximate-nearest-neighbor-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Approximate nearest neighbor search is widely used in vector search and dense retrieval systems, especially at large scale. It is not the same as vector search itself, because vector search may use exact search, approximate search, or other retrieval mechanisms.","status":"published","uri":"https://id.searchplex.net/approximate-nearest-neighbor-search/"}
{"artifactKind":"Dataset","definition":"An argument-retrieval evaluation dataset used to retrieve counterarguments or related argumentative texts.","distinctions":[{"reason":"ArguAna is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"ArguAna is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"}],"id":"arguana","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ArguAna in BEIR datasets","type":"officialHomepage","url":"https://github.com/beir-cellar/beir/wiki/Datasets-available"},{"label":"ArguAna in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/arguana"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"ArguAna","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"ArguAna is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/arguana/","json":"https://id.searchplex.net/arguana.json","jsonld":"https://id.searchplex.net/arguana.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"ArguAna is included in BEIR as an argument-retrieval dataset. It is a component evaluation dataset rather than a benchmark suite.","status":"published","uri":"https://id.searchplex.net/arguana/"}
{"altLabels":["AP"],"definition":"A ranking evaluation metric that averages precision values at the ranks where relevant items are retrieved for a single query or topic.","derivedRelations":[{"target":"ndcg","type":"related"}],"distinctions":[{"reason":"Average Precision is computed for one query or topic; MAP is the mean of AP over many queries or topics.","statement":"Average Precision is not Mean Average Precision.","target":"mean-average-precision","targetLabel":"Mean Average Precision","targetUri":"https://id.searchplex.net/mean-average-precision/"},{"reason":"Precision@K measures precision at a fixed cutoff; Average Precision aggregates precision at relevant ranks across the ranked list.","statement":"Average Precision is not Precision@K.","target":"precision-at-k","targetLabel":"Precision@K","targetUri":"https://id.searchplex.net/precision-at-k/"}],"id":"average-precision","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Average Precision","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"precision-metric","type":"broader"},{"target":"rank-sensitive-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"Average Precision is computed for one query or topic; MAP is the mean of AP over many queries or topics.","target":"mean-average-precision","type":"notEquivalentTo"},{"note":"Precision@K measures precision at a fixed cutoff; Average Precision aggregates precision at relevant ranks across the ranked list.","target":"precision-at-k","type":"notEquivalentTo"},{"target":"mean-average-precision","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"qrels","type":"related"},{"target":"recall-at-k","type":"related"}],"representations":{"html":"https://id.searchplex.net/average-precision/","json":"https://id.searchplex.net/average-precision.json","jsonld":"https://id.searchplex.net/average-precision.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Average Precision is a per-query or per-topic metric. Mean Average Precision averages AP values across multiple queries or topics.","status":"published","uri":"https://id.searchplex.net/average-precision/"}
{"artifactKind":"Benchmark","definition":"A benchmark suite for evaluating information-retrieval systems across multiple evaluation datasets, domains, and retrieval settings.","derivedRelations":[{"note":"ArguAna is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"arguana","type":"notEquivalentTo"},{"note":"BrowseComp evaluates browsing agents finding hard-to-locate web information; BEIR is a fixed information-retrieval benchmark suite.","target":"browsecomp","type":"notEquivalentTo"},{"note":"BrowseComp-Plus evaluates deep-research agents and retrieval methods over BrowseComp-derived tasks; BEIR is an information-retrieval benchmark suite.","target":"browsecomp-plus","type":"notEquivalentTo"},{"note":"A corpus is a collection of retrievable items; BEIR is a benchmark suite aggregating multiple evaluation datasets.","target":"corpus","type":"notEquivalentTo"},{"note":"FEVER is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"fever","type":"notEquivalentTo"},{"note":"FiQA-2018 is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"fiqa-2018","type":"notEquivalentTo"},{"note":"HotpotQA is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"hotpotqa","type":"notEquivalentTo"},{"note":"ir_datasets provides access to BEIR and many other datasets; BEIR is a benchmark suite.","target":"ir-datasets","type":"notEquivalentTo"},{"note":"MTEB evaluates text embeddings across many NLP and retrieval-adjacent tasks; BEIR is an information-retrieval benchmark suite.","target":"mteb","type":"notEquivalentTo"},{"note":"Natural Questions is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"natural-questions","type":"notEquivalentTo"},{"note":"NFCorpus is a component evaluation dataset; BEIR is the benchmark suite that includes NFCorpus among other datasets.","target":"nfcorpus","type":"notEquivalentTo"},{"note":"SciFact is a component evaluation dataset; BEIR is the benchmark suite that includes SciFact among other datasets.","target":"scifact","type":"notEquivalentTo"},{"note":"TREC is an evaluation program and workshop series; BEIR is a benchmark suite aggregating multiple retrieval datasets.","target":"trec","type":"notEquivalentTo"},{"note":"TREC-COVID is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"trec-covid","type":"notEquivalentTo"},{"note":"BEIR evaluates retrieval systems over fixed IR datasets; web browsing agent evaluation measures agents that search and browse the web or web-like corpora.","target":"web-browsing-agent-evaluation","type":"notEquivalentTo"}],"distinctions":[{"reason":"BEIR includes datasets with relevance judgments; qrels are judgment files, not the benchmark suite itself.","statement":"BEIR is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"BEIR is a benchmark suite that aggregates multiple IR evaluation datasets or test collections; a test collection is one evaluation-resource pattern within such suites.","statement":"BEIR is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"beir","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"BEIR benchmark repository","type":"officialHomepage","url":"https://github.com/beir-cellar/beir"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"BEIR","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"ranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"BEIR includes datasets with relevance judgments; qrels are judgment files, not the benchmark suite itself.","target":"qrels","type":"notEquivalentTo"},{"note":"BEIR is a benchmark suite that aggregates multiple IR evaluation datasets or test collections; a test collection is one evaluation-resource pattern within such suites.","target":"test-collection","type":"notEquivalentTo"},{"target":"arguana","type":"related"},{"target":"fever","type":"related"},{"target":"fiqa-2018","type":"related"},{"target":"hotpotqa","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"mteb","type":"related"},{"target":"natural-questions","type":"related"},{"target":"ndcg","type":"related"},{"target":"nfcorpus","type":"related"},{"target":"qrels","type":"related"},{"target":"ranking","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"retrieval","type":"related"},{"target":"scifact","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec-covid","type":"related"}],"representations":{"html":"https://id.searchplex.net/beir/","json":"https://id.searchplex.net/beir.json","jsonld":"https://id.searchplex.net/beir.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"BEIR is not itself a single retrieval dataset. It aggregates multiple test collections or IR evaluation datasets and standardizes their use for evaluating retrieval effectiveness, especially zero-shot generalization.","status":"published","terms":[{"community":"academic","label":"BEIR Dataset","usage":"informal"},{"community":"academic","label":"BEIR Benchmark","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/beir/"}
{"definition":"A WAND-family dynamic-pruning method that uses score upper bounds computed for blocks of postings to enable more aggressive skipping during top-k retrieval.","distinctions":[{"reason":"Block-Max WAND is a WAND-family refinement using block-level upper bounds; WAND is the broader dynamic-pruning method.","statement":"Block-Max WAND is not WAND.","target":"wand","targetLabel":"WAND","targetUri":"https://id.searchplex.net/wand/"}],"id":"block-max-wand","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Block-Max WAND paper","type":"definingReference","url":"https://doi.org/10.1145/2484028.2484077"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Block-Max WAND","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"dynamic-pruning","type":"broader"},{"note":"Block-Max WAND is a WAND-family refinement using block-level upper bounds; WAND is the broader dynamic-pruning method.","target":"wand","type":"notEquivalentTo"},{"target":"wand","type":"related"}],"representations":{"html":"https://id.searchplex.net/block-max-wand/","json":"https://id.searchplex.net/block-max-wand.json","jsonld":"https://id.searchplex.net/block-max-wand.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Block-Max WAND is a refinement of WAND-style dynamic pruning rather than a generic synonym for WAND.","status":"published","uri":"https://id.searchplex.net/block-max-wand/"}
{"altLabels":["Okapi BM25"],"definition":"A term-based ranking function that scores documents for a query using term frequency, inverse document frequency, and document-length normalization.","derivedRelations":[{"note":"BM25 scores documents using term statistics; exact match requires specified terms, values, or phrases to match exactly.","target":"exact-match","type":"notEquivalentTo"},{"target":"exact-match","type":"related"}],"distinctions":[{"reason":"BM25 is the standard BM25 scoring function; BM25F is a field-aware variant for structured or fielded documents.","statement":"BM25 is not BM25F.","target":"bm25f","targetLabel":"BM25F","targetUri":"https://id.searchplex.net/bm25f/"},{"reason":"BM25 is a ranking function; an inverted index is an index structure commonly used to retrieve postings for lexical scoring.","statement":"BM25 is not Inverted Index.","target":"inverted-index","targetLabel":"Inverted Index","targetUri":"https://id.searchplex.net/inverted-index/"},{"reason":"BM25 is a scoring method; keyword search is a practitioner-facing search style that may use BM25 or other lexical methods.","statement":"BM25 is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"BM25 is one lexical scoring method; lexical retrieval is the broader retrieval family based on lexical features.","statement":"BM25 is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"},{"reason":"BM25 uses sparse lexical evidence, but sparse retrieval also includes other traditional and learned sparse approaches.","statement":"BM25 is not Sparse Retrieval.","target":"sparse-retrieval","targetLabel":"Sparse Retrieval","targetUri":"https://id.searchplex.net/sparse-retrieval/"},{"reason":"BM25 and TF-IDF are both classical lexical weighting approaches, but BM25 adds probabilistic term saturation and document-length normalization.","statement":"BM25 is not TF-IDF.","target":"tf-idf","targetLabel":"TF-IDF","targetUri":"https://id.searchplex.net/tf-idf/"}],"id":"bm25","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"The Probabilistic Relevance Framework: BM25 and Beyond","type":"canonicalSurvey","url":"https://www.staff.city.ac.uk/~sbrp622/papers/foundations_bm25_review.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"BM25","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BM25 is the standard BM25 scoring function; BM25F is a field-aware variant for structured or fielded documents.","target":"bm25f","type":"notEquivalentTo"},{"note":"BM25 is a ranking function; an inverted index is an index structure commonly used to retrieve postings for lexical scoring.","target":"inverted-index","type":"notEquivalentTo"},{"note":"BM25 is a scoring method; keyword search is a practitioner-facing search style that may use BM25 or other lexical methods.","target":"keyword-search","type":"notEquivalentTo"},{"note":"BM25 is one lexical scoring method; lexical retrieval is the broader retrieval family based on lexical features.","target":"lexical-retrieval","type":"notEquivalentTo"},{"note":"BM25 uses sparse lexical evidence, but sparse retrieval also includes other traditional and learned sparse approaches.","target":"sparse-retrieval","type":"notEquivalentTo"},{"note":"BM25 and TF-IDF are both classical lexical weighting approaches, but BM25 adds probabilistic term saturation and document-length normalization.","target":"tf-idf","type":"notEquivalentTo"},{"target":"bm25f","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"inverted-index","type":"related"},{"target":"keyword-search","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"sparse-retrieval","type":"related"},{"target":"tf-idf","type":"related"},{"target":"ranking","type":"usedFor"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/bm25/","json":"https://id.searchplex.net/bm25.json","jsonld":"https://id.searchplex.net/bm25.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"BM25 is a widely used lexical retrieval and ranking method. It is often used as a baseline or component in keyword, lexical, sparse, and hybrid retrieval systems, but it is not itself the same as keyword search, lexical retrieval, or sparse retrieval.","status":"published","uri":"https://id.searchplex.net/bm25/"}
{"definition":"A field-aware variant of BM25 that combines term evidence from multiple document fields with field-specific weights or normalization.","distinctions":[{"reason":"BM25F is a field-aware BM25-family method; BM25 usually refers to the standard non-fielded scoring function.","statement":"BM25F is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"BM25F is one field-aware lexical scoring method; lexical retrieval is the broader retrieval family.","statement":"BM25F is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"}],"id":"bm25f","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Microsoft Cambridge at TREC-13: Web and HARD tracks","type":"definingReference","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/sigir-2004-mapinference.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"BM25F","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BM25F is a field-aware BM25-family method; BM25 usually refers to the standard non-fielded scoring function.","target":"bm25","type":"notEquivalentTo"},{"note":"BM25F is one field-aware lexical scoring method; lexical retrieval is the broader retrieval family.","target":"lexical-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"tf-idf","type":"related"},{"target":"ranking","type":"usedFor"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/bm25f/","json":"https://id.searchplex.net/bm25f.json","jsonld":"https://id.searchplex.net/bm25f.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"BM25F is used for fielded documents such as title, body, metadata, or anchor text. It is a BM25-family method, not a synonym for BM25 generally or for fielded retrieval as a whole.","status":"published","uri":"https://id.searchplex.net/bm25f/"}
{"altLabels":["Boolean Search"],"definition":"A retrieval method in which queries are expressed as Boolean combinations of terms or conditions and matching items satisfy the Boolean expression.","distinctions":[{"reason":"Exact match is a matching constraint; Boolean retrieval is a query model using Boolean operators over terms or conditions.","statement":"Boolean Retrieval is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"Boolean retrieval uses logical query expressions; keyword search is a broader practitioner-facing search style.","statement":"Boolean Retrieval is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"Boolean retrieval is based on satisfying logical conditions; ranking orders results by estimated relevance or utility.","statement":"Boolean Retrieval is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"}],"id":"boolean-retrieval","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Boolean retrieval","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/boolean-retrieval-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Boolean Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Exact match is a matching constraint; Boolean retrieval is a query model using Boolean operators over terms or conditions.","target":"exact-match","type":"notEquivalentTo"},{"note":"Boolean retrieval uses logical query expressions; keyword search is a broader practitioner-facing search style.","target":"keyword-search","type":"notEquivalentTo"},{"note":"Boolean retrieval is based on satisfying logical conditions; ranking orders results by estimated relevance or utility.","target":"ranking","type":"notEquivalentTo"},{"target":"exact-match","type":"related"},{"target":"inverted-index","type":"related"},{"target":"keyword-search","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/boolean-retrieval/","json":"https://id.searchplex.net/boolean-retrieval.json","jsonld":"https://id.searchplex.net/boolean-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Boolean retrieval returns matches according to logical conditions rather than necessarily producing a relevance-ranked list. It is distinct from keyword search, exact match, and ranked lexical retrieval, although systems may combine these ideas.","status":"published","uri":"https://id.searchplex.net/boolean-retrieval/"}
{"artifactKind":"Benchmark","definition":"A benchmark for evaluating browsing agents on their ability to locate hard-to-find information on the web and produce short, verifiable answers.","derivedRelations":[{"note":"BrowseComp is a benchmark for evaluating browsing agents; a web browsing agent is the system being evaluated.","target":"web-browsing-agent","type":"notEquivalentTo"}],"distinctions":[{"reason":"BrowseComp evaluates browsing agents finding hard-to-locate web information; BEIR is a fixed information-retrieval benchmark suite.","statement":"BrowseComp is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"BrowseComp evaluates browsing agents in live web conditions; BrowseComp-Plus derives from BrowseComp but uses a fixed curated corpus, supporting documents, and hard negatives for controlled evaluation.","statement":"BrowseComp is not BrowseComp-Plus.","target":"browsecomp-plus","targetLabel":"BrowseComp-Plus","targetUri":"https://id.searchplex.net/browsecomp-plus/"},{"reason":"BrowseComp is a browsing-agent benchmark; a test collection is a fixed IR evaluation resource consisting of corpus, topics or queries, and relevance judgments.","statement":"BrowseComp is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"browsecomp","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"BrowseComp","type":"officialHomepage","url":"https://openai.com/index/browsecomp/"},{"label":"BrowseComp paper","type":"definingReference","url":"https://arxiv.org/abs/2504.12516"},{"label":"OpenAI simple-evals","type":"officialRepository","url":"https://github.com/openai/simple-evals"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"BrowseComp","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"deep-research","type":"evaluates"},{"target":"web-browsing-agent","type":"evaluates"},{"target":"web-browsing-agent-evaluation","type":"evaluates"},{"note":"BrowseComp evaluates browsing agents finding hard-to-locate web information; BEIR is a fixed information-retrieval benchmark suite.","target":"beir","type":"notEquivalentTo"},{"note":"BrowseComp evaluates browsing agents in live web conditions; BrowseComp-Plus derives from BrowseComp but uses a fixed curated corpus, supporting documents, and hard negatives for controlled evaluation.","target":"browsecomp-plus","type":"notEquivalentTo"},{"note":"BrowseComp is a browsing-agent benchmark; a test collection is a fixed IR evaluation resource consisting of corpus, topics or queries, and relevance judgments.","target":"test-collection","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"deep-research","type":"related"},{"target":"web-browsing-agent","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"},{"target":"web-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/browsecomp/","json":"https://id.searchplex.net/browsecomp.json","jsonld":"https://id.searchplex.net/browsecomp.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"BrowseComp evaluates web-browsing agent behavior over live web search and browsing conditions. It is not a fixed IR test collection in the BEIR sense and should be distinguished from BrowseComp-Plus, which introduces a fixed curated corpus for more controlled evaluation.","status":"published","uri":"https://id.searchplex.net/browsecomp/"}
{"altLabels":["BrowseComp Plus"],"artifactKind":"Benchmark","definition":"A controlled benchmark for evaluating deep-research agents and retrievers, derived from BrowseComp and built around a fixed curated corpus with supporting documents and hard negatives.","distinctions":[{"reason":"BrowseComp-Plus evaluates deep-research agents and retrieval methods over BrowseComp-derived tasks; BEIR is an information-retrieval benchmark suite.","statement":"BrowseComp-Plus is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"BrowseComp-Plus derives from BrowseComp but uses a fixed curated corpus and supporting evidence; BrowseComp evaluates browsing agents under live web browsing conditions.","statement":"BrowseComp-Plus is not BrowseComp.","target":"browsecomp","targetLabel":"BrowseComp","targetUri":"https://id.searchplex.net/browsecomp/"},{"reason":"BrowseComp-Plus evaluates deep-research agents and retrievers; MTEB evaluates text embedding models across multiple tasks.","statement":"BrowseComp-Plus is not MTEB.","target":"mteb","targetLabel":"MTEB","targetUri":"https://id.searchplex.net/mteb/"}],"id":"browsecomp-plus","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"BrowseComp-Plus","type":"definingReference","url":"https://arxiv.org/abs/2508.06600"},{"label":"BrowseComp-Plus repository","type":"officialRepository","url":"https://github.com/Tevatron/BrowseComp-Plus"},{"label":"BrowseComp-Plus on Hugging Face","type":"officialHomepage","url":"https://huggingface.co/datasets/Tevatron/browsecomp-plus"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"BrowseComp-Plus","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"agentic-retrieval","type":"evaluates"},{"target":"deep-research","type":"evaluates"},{"target":"web-browsing-agent-evaluation","type":"evaluates"},{"note":"BrowseComp-Plus evaluates deep-research agents and retrieval methods over BrowseComp-derived tasks; BEIR is an information-retrieval benchmark suite.","target":"beir","type":"notEquivalentTo"},{"note":"BrowseComp-Plus derives from BrowseComp but uses a fixed curated corpus and supporting evidence; BrowseComp evaluates browsing agents under live web browsing conditions.","target":"browsecomp","type":"notEquivalentTo"},{"note":"BrowseComp-Plus evaluates deep-research agents and retrievers; MTEB evaluates text embedding models across multiple tasks.","target":"mteb","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"browsecomp","type":"related"},{"target":"corpus","type":"related"},{"target":"deep-research","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"retrieval-planning","type":"related"},{"target":"test-collection","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"}],"representations":{"html":"https://id.searchplex.net/browsecomp-plus/","json":"https://id.searchplex.net/browsecomp-plus.json","jsonld":"https://id.searchplex.net/browsecomp-plus.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"BrowseComp-Plus is designed to make evaluation of deep-research agents and retrieval methods more fair and transparent by controlling the document corpus and evidence. It is related to BrowseComp but should not be treated as identical to live-web browsing evaluation.","status":"published","terms":[{"community":"academic","label":"BCP","usage":"abbreviation"}],"uri":"https://id.searchplex.net/browsecomp-plus/"}
{"altLabels":["Candidate Selection"],"definition":"Retrieving an initial set of candidate items for later ranking, filtering, or reranking.","derivedRelations":[{"target":"recall-at-k","type":"evaluatedBy"},{"target":"chunking","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"ranking","type":"related"},{"target":"recall-metric","type":"related"},{"target":"result-filtering","type":"related"},{"target":"retrieval","type":"related"},{"target":"dense-passage-retrieval","type":"uses"},{"target":"dual-encoder","type":"uses"},{"target":"splade","type":"uses"},{"target":"weakand","type":"uses"}],"distinctions":[{"reason":"Candidate retrieval selects an initial set; first-stage ranking produces an ordered result list.","statement":"Candidate Retrieval is not First-Stage Ranking.","target":"first-stage-ranking","targetLabel":"First-Stage Ranking","targetUri":"https://id.searchplex.net/first-stage-ranking/"}],"id":"candidate-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Retrieve & Re-Rank","type":"officialHomepage","url":"https://www.sbert.net/examples/applications/retrieve_rerank/README.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Candidate Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Candidate retrieval selects an initial set; first-stage ranking produces an ordered result list.","target":"first-stage-ranking","type":"notEquivalentTo"},{"target":"retrieve-and-rerank","type":"partOf"},{"note":"First-stage ranking often operates over retrieved candidates.","target":"first-stage-ranking","type":"related"},{"target":"multi-stage-ranking","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"reranking","type":"related"}],"representations":{"html":"https://id.searchplex.net/candidate-retrieval/","json":"https://id.searchplex.net/candidate-retrieval.json","jsonld":"https://id.searchplex.net/candidate-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"The boundary with first-stage retrieval is editorially sensitive and should be reviewed per use case.","status":"published","terms":[{"community":"industry","label":"Candidate Generation","usage":"communityPreferred"},{"community":"academic","label":"First-Stage Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/candidate-retrieval/"}
{"definition":"A method for splitting larger content into smaller units that can be indexed, retrieved, embedded, passed to a model, or otherwise processed.","derivedRelations":[{"note":"Query segmentation splits a short query into meaningful spans; chunking usually splits larger content into retrievable or processable units.","target":"query-segmentation","type":"notEquivalentTo"},{"target":"query-segmentation","type":"related"}],"distinctions":[{"reason":"Chunking decides the units to represent; embedding is the representation of a unit.","statement":"Chunking is not Embedding.","target":"embedding","targetLabel":"Embedding","targetUri":"https://id.searchplex.net/embedding/"},{"reason":"Chunking creates passage-like units; passage retrieval retrieves such units for a query or information need.","statement":"Chunking is not Passage Retrieval.","target":"passage-retrieval","targetLabel":"Passage Retrieval","targetUri":"https://id.searchplex.net/passage-retrieval/"},{"reason":"Chunking is one design choice inside many RAG systems; RAG is the broader retrieval-augmented generation pattern.","statement":"Chunking is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"},{"reason":"Chunking creates or selects retrievable units; retrieval selects relevant units from an indexed collection.","statement":"Chunking is not Retrieval.","target":"retrieval","targetLabel":"Retrieval","targetUri":"https://id.searchplex.net/retrieval/"}],"id":"chunking","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Chunking strategies for LLM applications","type":"officialHomepage","url":"https://www.pinecone.io/learn/chunking-strategies/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Chunking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Chunking decides the units to represent; embedding is the representation of a unit.","target":"embedding","type":"notEquivalentTo"},{"note":"Chunking creates passage-like units; passage retrieval retrieves such units for a query or information need.","target":"passage-retrieval","type":"notEquivalentTo"},{"note":"Chunking is one design choice inside many RAG systems; RAG is the broader retrieval-augmented generation pattern.","target":"rag","type":"notEquivalentTo"},{"note":"Chunking creates or selects retrievable units; retrieval selects relevant units from an indexed collection.","target":"retrieval","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"embedding","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"rag","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/chunking/","json":"https://id.searchplex.net/chunking.json","jsonld":"https://id.searchplex.net/chunking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In RAG and search systems, chunking defines candidate units such as passages, sections, windows, or semantic segments. It influences retrieval quality but is not itself retrieval, ranking, or embedding.","status":"published","uri":"https://id.searchplex.net/chunking/"}
{"altLabels":["Conference and Labs of the Evaluation Forum","Cross-Language Evaluation Forum"],"definition":"The Conference and Labs of the Evaluation Forum, an evaluation program and research venue for information-access systems, especially multilingual and multimodal information access.","derivedRelations":[{"note":"FIRE and CLEF are distinct evaluation programs and communities, even though both support multilingual information-access evaluation.","target":"fire","type":"notEquivalentTo"},{"note":"INEX is a structured-document retrieval evaluation initiative that later ran within CLEF; CLEF is the broader evaluation forum.","target":"inex","type":"notEquivalentTo"},{"note":"NTCIR and CLEF are distinct evaluation programs with different communities, histories, and task emphases.","target":"ntcir","type":"notEquivalentTo"},{"target":"fire","type":"related"},{"target":"inex","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"ntcir","type":"related"}],"distinctions":[{"reason":"CLEF has historically emphasized cross-language and multilingual information access evaluation; cross-language retrieval is a retrieval task, not the CLEF venue.","statement":"CLEF is not Cross-Language Retrieval.","target":"cross-language-retrieval","targetLabel":"Cross-Language Retrieval","targetUri":"https://id.searchplex.net/cross-language-retrieval/"},{"reason":"CLEF labs may publish relevance judgments or qrels; qrels are judgment files, not the CLEF program itself.","statement":"CLEF is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"CLEF is an evaluation program and research venue; a test collection is an evaluation resource, some of which are created through CLEF labs.","statement":"CLEF is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"},{"reason":"CLEF and TREC are distinct evaluation programs and venues, though both support shared evaluation in information retrieval and information access.","statement":"CLEF is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"clef","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"CLEF Initiative","type":"officialHomepage","url":"https://www.clef-initiative.eu/"},{"label":"CLEF 2024","type":"officialHomepage","url":"https://clef2024.imag.fr/"},{"label":"Information Retrieval Evaluation in a Changing World","type":"canonicalSurvey","url":"https://doi.org/10.1007/978-3-030-22948-1"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"CLEF","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"CLEF has historically emphasized cross-language and multilingual information access evaluation; cross-language retrieval is a retrieval task, not the CLEF venue.","target":"cross-language-retrieval","type":"notEquivalentTo"},{"note":"CLEF labs may publish relevance judgments or qrels; qrels are judgment files, not the CLEF program itself.","target":"qrels","type":"notEquivalentTo"},{"note":"CLEF is an evaluation program and research venue; a test collection is an evaluation resource, some of which are created through CLEF labs.","target":"test-collection","type":"notEquivalentTo"},{"note":"CLEF and TREC are distinct evaluation programs and venues, though both support shared evaluation in information retrieval and information access.","target":"trec","type":"notEquivalentTo"},{"target":"cross-language-retrieval","type":"related"},{"target":"multilingual-retrieval","type":"related"},{"target":"multimodal-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"question-answering","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/clef/","json":"https://id.searchplex.net/clef.json","jsonld":"https://id.searchplex.net/clef.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"CLEF organizes evaluation labs, shared tasks, workshops, conference publications, and reusable test collections. It is not itself a metric, dataset, qrels file, or a single retrieval task, though CLEF labs may define tasks and publish datasets, runs, and relevance judgments.","status":"published","terms":[{"community":"academic","label":"CLEF Initiative","usage":"communityPreferred"},{"community":"academic","label":"CLEF Labs","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/clef/"}
{"artifactKind":"Model","definition":"A neural retrieval model that represents queries and documents using contextualized token-level vectors and scores them using late interaction.","derivedRelations":[{"note":"MaxSim is a scoring operation; ColBERT is a model architecture that uses MaxSim-style late interaction.","target":"maxsim","type":"notEquivalentTo"},{"note":"Multi-vector retrieval is the broader representation pattern; ColBERT is a specific late-interaction model family.","target":"multi-vector-retrieval","type":"notEquivalentTo"},{"note":"PLAID is a serving and retrieval engine; ColBERT is a late-interaction model family.","target":"plaid","type":"notEquivalentTo"}],"id":"colbert","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ColBERT paper","type":"definingReference","url":"https://doi.org/10.1145/3397271.3401075"},{"label":"ColBERT repository","type":"officialRepository","url":"https://github.com/stanford-futuredata/ColBERT"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"ColBERT","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"late-interaction","type":"realizes"},{"target":"colbertv2","type":"related"},{"target":"maxsim","type":"related"},{"target":"multi-vector-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/colbert/","json":"https://id.searchplex.net/colbert.json","jsonld":"https://id.searchplex.net/colbert.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"ColBERT is a concrete model architecture that introduced a widely used form of contextualized late interaction. It is one realization of the broader late-interaction paradigm, not a synonym for every late-interaction model.","status":"published","uri":"https://id.searchplex.net/colbert/"}
{"artifactKind":"Model","definition":"A refinement of ColBERT that improves the effectiveness and storage efficiency of late-interaction retrieval through improved supervision and compressed representations.","distinctions":[{"reason":"ColBERTv2 is a late-interaction model; PLAID is a retrieval engine and optimization for serving late-interaction representations.","statement":"ColBERTv2 is not PLAID.","target":"plaid","targetLabel":"PLAID","targetUri":"https://id.searchplex.net/plaid/"}],"id":"colbertv2","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ColBERTv2 paper","type":"definingReference","url":"https://doi.org/10.18653/v1/2022.naacl-main.272"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"ColBERTv2","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"ColBERTv2 is a late-interaction model; PLAID is a retrieval engine and optimization for serving late-interaction representations.","target":"plaid","type":"notEquivalentTo"},{"target":"late-interaction","type":"realizes"},{"target":"colbert","type":"related"},{"target":"maxsim","type":"related"},{"target":"plaid","type":"related"}],"representations":{"html":"https://id.searchplex.net/colbertv2/","json":"https://id.searchplex.net/colbertv2.json","jsonld":"https://id.searchplex.net/colbertv2.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"ColBERTv2 remains a ColBERT-family late-interaction retriever. It is distinct from PLAID, which is a retrieval engine and optimization for serving late-interaction models.","status":"published","uri":"https://id.searchplex.net/colbertv2/"}
{"altLabels":["Conversational Information Retrieval","CIR"],"definition":"A search task or interaction pattern in which users seek information through a multi-turn natural-language conversation with a search or retrieval system.","derivedRelations":[{"note":"Conversational search is a dialogue-oriented form of interactive search; interactive search also includes non-conversational interactions such as faceting, browsing, and reformulation.","target":"interactive-search","type":"notEquivalentTo"},{"note":"Session search uses session context across interactions; conversational search specifically involves a conversational interface or dialogue.","target":"session-search","type":"notEquivalentTo"}],"distinctions":[{"reason":"Conversational search centers on user-system dialogue; agentic retrieval centers on an agent controlling retrieval actions.","statement":"Conversational Search is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Conversational search is a multi-turn search interaction pattern; RAG is a generation pattern using retrieved information.","statement":"Conversational Search is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"conversational-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Neural Approaches to Conversational Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2201.05176"},{"label":"Conversational Search report","type":"canonicalSurvey","url":"https://arxiv.org/abs/2005.08658"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Conversational Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"interactive-search","type":"broader"},{"note":"Conversational search centers on user-system dialogue; agentic retrieval centers on an agent controlling retrieval actions.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Conversational search is a multi-turn search interaction pattern; RAG is a generation pattern using retrieved information.","target":"rag","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"rag","type":"related"},{"target":"session-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/conversational-search/","json":"https://id.searchplex.net/conversational-search.json","jsonld":"https://id.searchplex.net/conversational-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Conversational search emphasizes dialogue, context carryover, clarification, and multi-turn interaction. It overlaps with conversational information retrieval and interactive search, but is not simply a single query or a RAG system.","status":"published","uri":"https://id.searchplex.net/conversational-search/"}
{"altLabels":["Document Corpus","Document Collection"],"artifactKind":"Dataset","definition":"A collection of documents, passages, web pages, items, or other retrievable units used for search, retrieval, training, or evaluation.","derivedRelations":[{"target":"browsecomp-plus","type":"related"}],"distinctions":[{"reason":"A corpus is a collection of retrievable items; BEIR is a benchmark suite aggregating multiple evaluation datasets.","statement":"Corpus is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"A corpus contains retrievable items; qrels contain relevance judgments for query-item pairs.","statement":"Corpus is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"A corpus is the searchable collection; a test collection also includes queries or topics and relevance judgments.","statement":"Corpus is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"corpus","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An example information retrieval problem","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html"},{"label":"ir_datasets catalog","type":"officialHomepage","url":"https://ir-datasets.com/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Corpus","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"A corpus is a collection of retrievable items; BEIR is a benchmark suite aggregating multiple evaluation datasets.","target":"beir","type":"notEquivalentTo"},{"note":"A corpus contains retrievable items; qrels contain relevance judgments for query-item pairs.","target":"qrels","type":"notEquivalentTo"},{"note":"A corpus is the searchable collection; a test collection also includes queries or topics and relevance judgments.","target":"test-collection","type":"notEquivalentTo"},{"target":"test-collection","type":"partOf"},{"target":"document-retrieval","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"qrels","type":"related"}],"representations":{"html":"https://id.searchplex.net/corpus/","json":"https://id.searchplex.net/corpus.json","jsonld":"https://id.searchplex.net/corpus.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In IR evaluation, the corpus is the searchable collection component of a test collection. The bare term collection is ambiguous: it may mean a corpus or searchable document collection, while test collection usually means the full evaluation package with corpus, topics or queries, and relevance judgments.","status":"published","terms":[{"community":"academic","label":"Collection","usage":"nearSynonym"},{"community":"academic","label":"Search Collection","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/corpus/"}
{"definition":"A neural architecture that jointly processes both sides of a matching problem, such as a query and a document, allowing direct interaction between them before producing a relevance, similarity, or classification score.","derivedRelations":[{"note":"A dense encoder produces a dense vector for an input; a cross-encoder jointly processes an input pair.","target":"dense-encoder","type":"notEquivalentTo"},{"note":"Retrieve-and-rerank is a pipeline pattern; a Cross-Encoder is one possible reranking method.","target":"retrieve-and-rerank","type":"notEquivalentTo"}],"distinctions":[{"reason":"Cross-encoders process a pair jointly; dual encoders encode each side independently and compare representations.","statement":"Cross-Encoder is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"},{"reason":"Cross-Encoder names a model architecture or method; ranking is the task of ordering results.","statement":"Cross-Encoder is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"A Cross-Encoder is one common way to perform reranking, but reranking is the retrieval-system task or stage.","statement":"Cross-Encoder is not Reranking.","target":"reranking","targetLabel":"Reranking","targetUri":"https://id.searchplex.net/reranking/"}],"id":"cross-encoder","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Sentence Transformers CrossEncoder documentation","type":"officialHomepage","url":"https://www.sbert.net/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Cross-Encoder","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Cross-encoders process a pair jointly; dual encoders encode each side independently and compare representations.","target":"dual-encoder","type":"notEquivalentTo"},{"note":"Cross-Encoder names a model architecture or method; ranking is the task of ordering results.","target":"ranking","type":"notEquivalentTo"},{"note":"A Cross-Encoder is one common way to perform reranking, but reranking is the retrieval-system task or stage.","target":"reranking","type":"notEquivalentTo"},{"target":"dual-encoder","type":"related"},{"target":"multi-stage-ranking","type":"related"},{"target":"retrieve-and-rerank","type":"related"},{"target":"ranking","type":"usedFor"},{"target":"reranking","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/cross-encoder/","json":"https://id.searchplex.net/cross-encoder.json","jsonld":"https://id.searchplex.net/cross-encoder.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Because each query-candidate pair must be processed jointly, Cross-Encoders are usually much more expensive for large-scale retrieval than independently encoded representations. In search systems they are therefore commonly used to rerank a bounded candidate set.","status":"published","terms":[{"community":"industry","label":"Reranker","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/cross-encoder/"}
{"altLabels":["Cross-Language Information Retrieval","Cross-Lingual Retrieval","Cross-Lingual Information Retrieval","CLIR"],"definition":"Retrieval in which the user's query or information need is expressed in one language and relevant retrieved items may be written in another language.","derivedRelations":[{"note":"CLEF has historically emphasized cross-language and multilingual information access evaluation; cross-language retrieval is a retrieval task, not the CLEF venue.","target":"clef","type":"notEquivalentTo"},{"target":"fire","type":"related"},{"target":"ntcir","type":"related"}],"distinctions":[{"reason":"Cross-language retrieval specifically crosses a query-result language boundary; multilingual retrieval also covers broader multi-language collections and systems.","statement":"Cross-Language Retrieval is not Multilingual Retrieval.","target":"multilingual-retrieval","targetLabel":"Multilingual Retrieval","targetUri":"https://id.searchplex.net/multilingual-retrieval/"}],"id":"cross-language-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Cross-language Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2111.05988"},{"label":"Text REtrieval Conference","type":"officialHomepage","url":"https://trec.nist.gov/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Cross-Language Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"multilingual-retrieval","type":"broader"},{"note":"Cross-language retrieval specifically crosses a query-result language boundary; multilingual retrieval also covers broader multi-language collections and systems.","target":"multilingual-retrieval","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/cross-language-retrieval/","json":"https://id.searchplex.net/cross-language-retrieval.json","jsonld":"https://id.searchplex.net/cross-language-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Cross-language retrieval is often called cross-lingual information retrieval. It is narrower than multilingual retrieval and commonly involves query translation, document translation, multilingual representations, or cross-lingual embeddings.","status":"published","uri":"https://id.searchplex.net/cross-language-retrieval/"}
{"altLabels":["Cross-Modality Retrieval","Cross-Modal Search"],"definition":"A retrieval task in which the query and retrieved items may be expressed in different modalities, such as retrieving images from text queries or text from image queries.","derivedRelations":[{"note":"Multimodal retrieval is the broader family involving multiple modalities; cross-modal retrieval specifically crosses a modality boundary between query and retrieved items.","target":"multimodal-retrieval","type":"notEquivalentTo"},{"target":"trecvid","type":"related"}],"distinctions":[{"reason":"Image retrieval retrieves images; cross-modal retrieval specifically matches across modalities, such as text-to-image or image-to-text.","statement":"Cross-Modal Retrieval is not Image Retrieval.","target":"image-retrieval","targetLabel":"Image Retrieval","targetUri":"https://id.searchplex.net/image-retrieval/"},{"reason":"Reverse image search uses an image as the query, often to find visually similar or related images; cross-modal retrieval crosses between different modalities.","statement":"Cross-Modal Retrieval is not Reverse Image Search.","target":"reverse-image-search","targetLabel":"Reverse Image Search","targetUri":"https://id.searchplex.net/reverse-image-search/"},{"reason":"Cross-modal retrieval may use vector search, but it is a task involving modality boundaries rather than a vector-search mechanism itself.","statement":"Cross-Modal Retrieval is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"cross-modal-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"A Comprehensive Survey on Cross-modal Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/1607.06215"},{"label":"Image-text Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2203.14713"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Cross-Modal Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"multimodal-retrieval","type":"broader"},{"note":"Image retrieval retrieves images; cross-modal retrieval specifically matches across modalities, such as text-to-image or image-to-text.","target":"image-retrieval","type":"notEquivalentTo"},{"note":"Reverse image search uses an image as the query, often to find visually similar or related images; cross-modal retrieval crosses between different modalities.","target":"reverse-image-search","type":"notEquivalentTo"},{"note":"Cross-modal retrieval may use vector search, but it is a task involving modality boundaries rather than a vector-search mechanism itself.","target":"vector-search","type":"notEquivalentTo"},{"target":"embedding","type":"related"},{"target":"image-retrieval","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/cross-modal-retrieval/","json":"https://id.searchplex.net/cross-modal-retrieval.json","jsonld":"https://id.searchplex.net/cross-modal-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Cross-modal retrieval is a narrower case within multimodal retrieval. It emphasizes matching across a modality boundary rather than searching within one modality or combining multiple modalities in a single representation.","status":"published","terms":[{"community":"academic","label":"Image-Text Retrieval","usage":"nearSynonym"},{"community":"industry","label":"Text-to-Image Search","usage":"nearSynonym"},{"community":"industry","label":"Image-to-Text Search","usage":"nearSynonym"},{"community":"vendor","label":"Cross-Type Search","usage":"informal"}],"uri":"https://id.searchplex.net/cross-modal-retrieval/"}
{"altLabels":["At-K Metric","Top-K Metric","Metric@K"],"definition":"A family of evaluation metrics computed over only the top K retrieved or ranked results.","derivedRelations":[{"target":"ndcg","type":"narrower"},{"target":"precision-at-k","type":"narrower"},{"target":"recall-at-k","type":"narrower"},{"target":"success-at-k","type":"narrower"}],"id":"cutoff-metric","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Cutoff Metric","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"precision-metric","type":"related"},{"target":"ranking","type":"related"},{"target":"recall-metric","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/cutoff-metric/","json":"https://id.searchplex.net/cutoff-metric.json","jsonld":"https://id.searchplex.net/cutoff-metric.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Cutoff metrics are commonly written with an @K suffix, such as Precision@K, Recall@K, Success@K, or nDCG@K. They evaluate effectiveness at a user- or system-relevant rank cutoff rather than over an unbounded result list.","status":"published","uri":"https://id.searchplex.net/cutoff-metric/"}
{"altLabels":["Agentic Deep Research"],"definition":"An agentic information-seeking pattern in which a system performs multi-step research by planning searches, browsing or retrieving sources, reasoning over evidence, and synthesizing a cited answer or report.","derivedRelations":[{"target":"browsecomp","type":"evaluatedBy"},{"target":"browsecomp-plus","type":"evaluatedBy"},{"target":"browsecomp","type":"related"},{"target":"browsecomp-plus","type":"related"}],"distinctions":[{"reason":"Agentic RAG is a generation pattern with agentic retrieval or tool-use control; deep research emphasizes broader multi-step research and cited synthesis.","statement":"Deep Research is not Agentic RAG.","target":"agentic-rag","targetLabel":"Agentic RAG","targetUri":"https://id.searchplex.net/agentic-rag/"},{"reason":"Deep research is a broader agentic research and synthesis pattern; agentic retrieval is the retrieval loop that may support it.","statement":"Deep Research is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Question answering returns an answer to a question; deep research often produces a synthesized, cited research result after multiple search and reasoning steps.","statement":"Deep Research is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"},{"reason":"RAG is a retrieval-augmented generation pattern; deep research emphasizes multi-step search, browsing, evidence gathering, and cited synthesis.","statement":"Deep Research is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"deep-research","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Introducing deep research","type":"officialHomepage","url":"https://openai.com/index/introducing-deep-research/"},{"label":"From Web Search towards Agentic Deep Research","type":"canonicalSurvey","url":"https://arxiv.org/abs/2506.18959"},{"label":"DeepResearch Bench","type":"definingReference","url":"https://arxiv.org/abs/2506.11763"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Deep Research","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Agentic RAG is a generation pattern with agentic retrieval or tool-use control; deep research emphasizes broader multi-step research and cited synthesis.","target":"agentic-rag","type":"notEquivalentTo"},{"note":"Deep research is a broader agentic research and synthesis pattern; agentic retrieval is the retrieval loop that may support it.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Question answering returns an answer to a question; deep research often produces a synthesized, cited research result after multiple search and reasoning steps.","target":"question-answering","type":"notEquivalentTo"},{"note":"RAG is a retrieval-augmented generation pattern; deep research emphasizes multi-step search, browsing, evidence gathering, and cited synthesis.","target":"rag","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"agentic-retrieval","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"question-answering","type":"related"},{"target":"rag","type":"related"},{"target":"retrieval-planning","type":"related"},{"target":"web-browsing-agent","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"}],"representations":{"html":"https://id.searchplex.net/deep-research/","json":"https://id.searchplex.net/deep-research.json","jsonld":"https://id.searchplex.net/deep-research.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Deep research usually combines agentic retrieval, web browsing, source evaluation, citation, and synthesis. It is related to RAG and question answering, but it emphasizes multi-step research behavior and report-style synthesis rather than a single retrieval or generation step.","status":"published","terms":[{"community":"industry","label":"Deep Research Agent","usage":"nearSynonym"},{"community":"academic","label":"DRA","usage":"abbreviation"}],"uri":"https://id.searchplex.net/deep-research/"}
{"definition":"An encoder that maps an input, such as a query, document, image, or other item, to a dense vector representation.","derivedRelations":[{"note":"A dense encoder is a component or model that produces dense representations; an embedding is the produced representation.","target":"embedding","type":"notEquivalentTo"},{"target":"embedding","type":"related"}],"distinctions":[{"reason":"A dense encoder produces a dense vector for an input; a cross-encoder jointly processes an input pair.","statement":"Dense Encoder is not Cross-Encoder.","target":"cross-encoder","targetLabel":"Cross-Encoder","targetUri":"https://id.searchplex.net/cross-encoder/"},{"reason":"Dense Encoder names an encoding component; dense retrieval names a retrieval approach using dense representations.","statement":"Dense Encoder is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"A dual encoder combines independently encoded sides of a matching problem; a dense encoder is one encoder component.","statement":"Dense Encoder is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"}],"id":"dense-encoder","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Sentence Transformers embedding model documentation","type":"officialHomepage","url":"https://www.sbert.net/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Dense Encoder","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"A dense encoder produces a dense vector for an input; a cross-encoder jointly processes an input pair.","target":"cross-encoder","type":"notEquivalentTo"},{"note":"Dense Encoder names an encoding component; dense retrieval names a retrieval approach using dense representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"A dual encoder combines independently encoded sides of a matching problem; a dense encoder is one encoder component.","target":"dual-encoder","type":"notEquivalentTo"},{"target":"dense-retrieval","type":"related"},{"target":"dual-encoder","type":"related"}],"representations":{"html":"https://id.searchplex.net/dense-encoder/","json":"https://id.searchplex.net/dense-encoder.json","jsonld":"https://id.searchplex.net/dense-encoder.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In dense retrieval, dense encoders commonly produce query and document vectors for similarity-based retrieval. A dense encoder is an encoding component, not by itself a complete retrieval architecture. A dual encoder typically contains or uses independently applied encoders, but not every dense encoder is part of a dual-encoder retrieval system.","status":"published","uri":"https://id.searchplex.net/dense-encoder/"}
{"altLabels":["DPR"],"definition":"A neural passage-retrieval method that represents questions and passages with dense vectors, commonly using a dual-encoder architecture, to retrieve candidate passages.","distinctions":[{"reason":"Dense Passage Retrieval is a specific dense retrieval method; dense retrieval is the broader retrieval approach using dense representations.","statement":"Dense Passage Retrieval is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Dense Passage Retrieval commonly uses a dual-encoder architecture, but DPR names a specific retrieval method rather than the architecture itself.","statement":"Dense Passage Retrieval is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"},{"reason":"Dense Passage Retrieval is a specific dense neural method; passage retrieval is the broader task of retrieving passages.","statement":"Dense Passage Retrieval is not Passage Retrieval.","target":"passage-retrieval","targetLabel":"Passage Retrieval","targetUri":"https://id.searchplex.net/passage-retrieval/"},{"reason":"Semantic search is a broad meaning-oriented search goal or product capability; Dense Passage Retrieval is one method that can support such systems.","statement":"Dense Passage Retrieval is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Vector search is a similarity-search mechanism over vectors; Dense Passage Retrieval is a passage-retrieval method that may use vector search.","statement":"Dense Passage Retrieval is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"dense-passage-retrieval","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Dense Passage Retrieval for Open-Domain Question Answering","type":"definingReference","url":"https://arxiv.org/abs/2004.04906"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Dense Passage Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Dense Passage Retrieval is a specific dense retrieval method; dense retrieval is the broader retrieval approach using dense representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Dense Passage Retrieval commonly uses a dual-encoder architecture, but DPR names a specific retrieval method rather than the architecture itself.","target":"dual-encoder","type":"notEquivalentTo"},{"note":"Dense Passage Retrieval is a specific dense neural method; passage retrieval is the broader task of retrieving passages.","target":"passage-retrieval","type":"notEquivalentTo"},{"note":"Semantic search is a broad meaning-oriented search goal or product capability; Dense Passage Retrieval is one method that can support such systems.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Vector search is a similarity-search mechanism over vectors; Dense Passage Retrieval is a passage-retrieval method that may use vector search.","target":"vector-search","type":"notEquivalentTo"},{"target":"dense-retrieval","type":"related"},{"target":"dual-encoder","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"semantic-search","type":"related"},{"target":"vector-search","type":"related"},{"target":"candidate-retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/dense-passage-retrieval/","json":"https://id.searchplex.net/dense-passage-retrieval.json","jsonld":"https://id.searchplex.net/dense-passage-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Dense Passage Retrieval is often abbreviated DPR and is associated with the open-domain question-answering method introduced under that name. It is not a synonym for dense retrieval generally, vector search, or semantic search.","slugAliases":["dpr"],"status":"published","uri":"https://id.searchplex.net/dense-passage-retrieval/"}
{"definition":"Retrieval using dense vector representations of queries and items, typically compared using a vector similarity or distance function.","derivedRelations":[{"target":"mteb","type":"evaluatedBy"},{"note":"Approximate nearest neighbor search is an efficiency method for neighbor lookup; dense retrieval is a retrieval approach using dense representations.","target":"approximate-nearest-neighbor-search","type":"notEquivalentTo"},{"note":"HNSW can serve dense retrieval systems, but dense retrieval is the retrieval approach using dense representations.","target":"hnsw","type":"notEquivalentTo"},{"note":"Hybrid retrieval commonly combines dense retrieval with lexical or sparse retrieval; it is not the same as dense retrieval alone.","target":"hybrid-retrieval","type":"notEquivalentTo"},{"note":"Dense retrieval is one possible retrieval component in RAG; RAG is the broader system pattern.","target":"rag","type":"notEquivalentTo"},{"note":"The classical vector space model usually uses sparse term-weighted vectors; dense retrieval uses learned dense representations.","target":"vector-space-model","type":"notEquivalentTo"},{"target":"mteb","type":"related"},{"target":"rag","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"distinctions":[{"reason":"Dense retrieval is a retrieval approach; a dense encoder is an encoding component that may be used within it.","statement":"Dense Retrieval is not Dense Encoder.","target":"dense-encoder","targetLabel":"Dense Encoder","targetUri":"https://id.searchplex.net/dense-encoder/"},{"reason":"Dense Passage Retrieval is a specific dense retrieval method for passage retrieval; dense retrieval is the broader approach using dense representations.","statement":"Dense Retrieval is not Dense Passage Retrieval.","target":"dense-passage-retrieval","targetLabel":"Dense Passage Retrieval","targetUri":"https://id.searchplex.net/dense-passage-retrieval/"},{"reason":"Dense retrieval is a retrieval approach using dense representations; a dual encoder is one common architecture for implementing it.","statement":"Dense Retrieval is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"},{"reason":"Dense retrieval is a retrieval approach using dense representations; an embedding is a representation used within or outside retrieval.","statement":"Dense Retrieval is not Embedding.","target":"embedding","targetLabel":"Embedding","targetUri":"https://id.searchplex.net/embedding/"},{"reason":"Semantic search is an intent-oriented product term and may use dense retrieval, sparse retrieval, hybrid retrieval, or other methods.","statement":"Dense Retrieval is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Dense retrieval describes the use of dense representations; single-vector retrieval describes representing each retrievable item with one vector.","statement":"Dense Retrieval is not Single-Vector Retrieval.","target":"single-vector-retrieval","targetLabel":"Single-Vector Retrieval","targetUri":"https://id.searchplex.net/single-vector-retrieval/"},{"reason":"Dense retrieval is a specific learned dense-vector retrieval approach; vector search is a broader practitioner term.","statement":"Dense Retrieval is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"dense-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Dense Passage Retrieval","type":"definingReference","url":"https://arxiv.org/abs/2004.04906"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Dense Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Dense retrieval is a retrieval approach; a dense encoder is an encoding component that may be used within it.","target":"dense-encoder","type":"notEquivalentTo"},{"note":"Dense Passage Retrieval is a specific dense retrieval method for passage retrieval; dense retrieval is the broader approach using dense representations.","target":"dense-passage-retrieval","type":"notEquivalentTo"},{"note":"Dense retrieval is a retrieval approach using dense representations; a dual encoder is one common architecture for implementing it.","target":"dual-encoder","type":"notEquivalentTo"},{"note":"Dense retrieval is a retrieval approach using dense representations; an embedding is a representation used within or outside retrieval.","target":"embedding","type":"notEquivalentTo"},{"note":"Semantic search is an intent-oriented product term and may use dense retrieval, sparse retrieval, hybrid retrieval, or other methods.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Dense retrieval describes the use of dense representations; single-vector retrieval describes representing each retrievable item with one vector.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"note":"Dense retrieval is a specific learned dense-vector retrieval approach; vector search is a broader practitioner term.","target":"vector-search","type":"notEquivalentTo"},{"target":"approximate-nearest-neighbor-search","type":"related"},{"target":"dense-encoder","type":"related"},{"target":"dense-passage-retrieval","type":"related"},{"target":"dual-encoder","type":"related"},{"target":"embedding","type":"related"},{"target":"hnsw","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"semantic-search","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/dense-retrieval/","json":"https://id.searchplex.net/dense-retrieval.json","jsonld":"https://id.searchplex.net/dense-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Dense retrieval normally refers to learned low- or moderate-dimensional dense embeddings. It is narrower than the broad practitioner term vector search.","status":"published","terms":[{"community":"industry","label":"Embedding Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/dense-retrieval/"}
{"altLabels":["Diversity Retrieval","Search Result Diversification","Diversified Search"],"definition":"A retrieval task or objective in which a result set should cover multiple aspects, intents, interpretations, or subtopics of an information need.","distinctions":[{"reason":"Diversified retrieval adds an explicit coverage or diversity objective; ranking generally orders results by relevance or utility and may or may not optimize diversity.","statement":"Diversified Retrieval is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"Result fusion combines ranked lists or result sets; diversified retrieval is an objective about covering aspects or intents in the returned results.","statement":"Diversified Retrieval is not Result Fusion.","target":"result-fusion","targetLabel":"Result Fusion","targetUri":"https://id.searchplex.net/result-fusion/"}],"id":"diversified-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Search Result Diversification","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000046"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Diversified Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Diversified retrieval adds an explicit coverage or diversity objective; ranking generally orders results by relevance or utility and may or may not optimize diversity.","target":"ranking","type":"notEquivalentTo"},{"note":"Result fusion combines ranked lists or result sets; diversified retrieval is an objective about covering aspects or intents in the returned results.","target":"result-fusion","type":"notEquivalentTo"},{"target":"query-understanding","type":"related"},{"target":"ranking","type":"related"},{"target":"serp","type":"related"}],"representations":{"html":"https://id.searchplex.net/diversified-retrieval/","json":"https://id.searchplex.net/diversified-retrieval.json","jsonld":"https://id.searchplex.net/diversified-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Diversified retrieval is especially important for ambiguous or multifaceted queries, where returning near-duplicate results for one interpretation may be less useful than covering several plausible intents. It is distinct from ordinary ranking by relevance alone.","status":"published","uri":"https://id.searchplex.net/diversified-retrieval/"}
{"altLabels":["Document Expansion by Query Prediction"],"definition":"Enriching a document representation with additional generated or selected text, terms, queries, or signals intended to improve retrieval matching.","distinctions":[{"reason":"Document expansion enriches the indexed document representation; query expansion enriches the query representation.","statement":"Document Expansion is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"}],"id":"document-expansion","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Document Expansion by Query Prediction","type":"definingReference","url":"https://arxiv.org/abs/1904.08375"},{"label":"docTTTTTquery","type":"officialRepository","url":"https://github.com/castorini/docTTTTTquery"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Document Expansion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Document expansion enriches the indexed document representation; query expansion enriches the query representation.","target":"query-expansion","type":"notEquivalentTo"},{"target":"lexical-retrieval","type":"related"},{"target":"query-expansion","type":"related"},{"target":"retrieval","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"representations":{"html":"https://id.searchplex.net/document-expansion/","json":"https://id.searchplex.net/document-expansion.json","jsonld":"https://id.searchplex.net/document-expansion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Document expansion modifies or augments the document side, often before indexing. It is distinct from query expansion, which modifies or augments the query side at search time or query-processing time.","status":"published","terms":[{"community":"academic","label":"Doc2Query","usage":"nearSynonym"},{"community":"academic","label":"docT5query","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/document-expansion/"}
{"altLabels":["Document Search"],"definition":"A retrieval task in which the retrievable units are documents rather than passages, entities, answers, or other smaller or more structured objects.","derivedRelations":[{"note":"INEX is an evaluation initiative; document retrieval is a retrieval task evaluated by some INEX-style resources.","target":"inex","type":"notEquivalentTo"},{"target":"arguana","type":"related"},{"target":"corpus","type":"related"},{"target":"fever","type":"related"},{"target":"inex","type":"related"},{"target":"nfcorpus","type":"related"},{"target":"scifact","type":"related"},{"target":"trec-covid","type":"related"}],"distinctions":[{"reason":"Document retrieval targets documents; entity search targets entities or entity-oriented information.","statement":"Document Retrieval is not Entity Search.","target":"entity-search","targetLabel":"Entity Search","targetUri":"https://id.searchplex.net/entity-search/"},{"reason":"Document retrieval returns documents as the primary retrievable units; passage retrieval returns passages or passage-like segments.","statement":"Document Retrieval is not Passage Retrieval.","target":"passage-retrieval","targetLabel":"Passage Retrieval","targetUri":"https://id.searchplex.net/passage-retrieval/"},{"reason":"Document retrieval returns documents for an information need; question answering aims to return answers, answer-bearing snippets, or generated responses.","statement":"Document Retrieval is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"}],"id":"document-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Overview of the TREC 2019 Deep Learning Track","type":"definingReference","url":"https://arxiv.org/abs/2003.07820"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Document Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Document retrieval targets documents; entity search targets entities or entity-oriented information.","target":"entity-search","type":"notEquivalentTo"},{"note":"Document retrieval returns documents as the primary retrievable units; passage retrieval returns passages or passage-like segments.","target":"passage-retrieval","type":"notEquivalentTo"},{"note":"Document retrieval returns documents for an information need; question answering aims to return answers, answer-bearing snippets, or generated responses.","target":"question-answering","type":"notEquivalentTo"},{"target":"ad-hoc-retrieval","type":"related"},{"target":"ranking","type":"related"},{"target":"web-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/document-retrieval/","json":"https://id.searchplex.net/document-retrieval.json","jsonld":"https://id.searchplex.net/document-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Document retrieval is a core IR task formulation and is often used in ad hoc retrieval and web-search evaluation. It is distinct from passage retrieval, which retrieves passage-like units, and from question answering, which aims to return answers rather than documents.","status":"published","terms":[{"community":"industry","label":"Enterprise Document Search","usage":"nearSynonym"},{"community":"industry","label":"Knowledge Base Search","usage":"informal"}],"uri":"https://id.searchplex.net/document-retrieval/"}
{"altLabels":["Bi-Encoder"],"definition":"A neural architecture that encodes the two sides of a matching problem independently, such as a query and a document, and compares their resulting representations using a similarity or scoring function.","derivedRelations":[{"note":"Dense Passage Retrieval commonly uses a dual-encoder architecture, but DPR names a specific retrieval method rather than the architecture itself.","target":"dense-passage-retrieval","type":"notEquivalentTo"},{"note":"Retrieve-and-rerank is a pipeline pattern; a dual encoder is one possible first-stage method.","target":"retrieve-and-rerank","type":"notEquivalentTo"},{"note":"Dual encoders commonly produce single-vector representations, but the architecture and the representation format are distinct concepts.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"target":"single-vector-retrieval","type":"related"}],"distinctions":[{"reason":"Dual encoders encode each side independently; cross-encoders process the pair jointly.","statement":"Dual Encoder is not Cross-Encoder.","target":"cross-encoder","targetLabel":"Cross-Encoder","targetUri":"https://id.searchplex.net/cross-encoder/"},{"reason":"A dense encoder maps one input to a dense vector; a dual encoder applies encoders to two sides of a matching problem and compares the representations.","statement":"Dual Encoder is not Dense Encoder.","target":"dense-encoder","targetLabel":"Dense Encoder","targetUri":"https://id.searchplex.net/dense-encoder/"},{"reason":"A dual encoder is an architecture or method; dense retrieval is retrieval using dense representations and is commonly implemented with dual encoders.","statement":"Dual Encoder is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"}],"id":"dual-encoder","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Sentence Transformers documentation","type":"officialHomepage","url":"https://www.sbert.net/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Dual Encoder","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Dual encoders encode each side independently; cross-encoders process the pair jointly.","target":"cross-encoder","type":"notEquivalentTo"},{"note":"A dense encoder maps one input to a dense vector; a dual encoder applies encoders to two sides of a matching problem and compares the representations.","target":"dense-encoder","type":"notEquivalentTo"},{"note":"A dual encoder is an architecture or method; dense retrieval is retrieval using dense representations and is commonly implemented with dual encoders.","target":"dense-retrieval","type":"notEquivalentTo"},{"target":"cross-encoder","type":"related"},{"target":"dense-encoder","type":"related"},{"target":"dense-passage-retrieval","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"retrieve-and-rerank","type":"related"},{"target":"candidate-retrieval","type":"usedFor"},{"target":"first-stage-ranking","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/dual-encoder/","json":"https://id.searchplex.net/dual-encoder.json","jsonld":"https://id.searchplex.net/dual-encoder.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In information retrieval, a dual encoder commonly uses separate query and document encoding paths so document representations can be computed ahead of time and efficiently searched. The two encoders may use separate parameters or share some or all parameters. Bi-encoder is commonly used as an equivalent term.","status":"published","terms":[{"community":"industry","label":"Two-Tower Model","usage":"communityPreferred"},{"community":"industry","label":"Two Tower","usage":"informal"}],"uri":"https://id.searchplex.net/dual-encoder/"}
{"definition":"A family of query-evaluation techniques that avoid fully evaluating candidates that cannot affect the desired top-ranked results.","derivedRelations":[{"target":"block-max-wand","type":"narrower"},{"target":"wand","type":"narrower"},{"target":"splade","type":"related"}],"id":"dynamic-pruning","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"WAND paper","type":"definingReference","url":"https://doi.org/10.1145/1571941.1572114"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Dynamic Pruning","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"learned-sparse-retrieval","type":"related"},{"target":"sparse-retrieval","type":"related"},{"target":"weakand","type":"related"}],"representations":{"html":"https://id.searchplex.net/dynamic-pruning/","json":"https://id.searchplex.net/dynamic-pruning.json","jsonld":"https://id.searchplex.net/dynamic-pruning.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Dynamic pruning uses score bounds or related statistics to skip unnecessary scoring work while preserving or approximating top-k effectiveness.","status":"published","uri":"https://id.searchplex.net/dynamic-pruning/"}
{"altLabels":["Vector Embedding"],"definition":"A vector representation of an item, token, query, document, or other object in a continuous or otherwise structured representation space.","derivedRelations":[{"target":"mteb","type":"evaluatedBy"},{"note":"Chunking decides the units to represent; embedding is the representation of a unit.","target":"chunking","type":"notEquivalentTo"},{"note":"MTEB is a benchmark suite for evaluating embedding models; an embedding is a vector representation.","target":"mteb","type":"notEquivalentTo"},{"note":"The vector space model is a retrieval model over term-weighted vectors; an embedding is a learned vector representation that may be used in many retrieval or non-retrieval settings.","target":"vector-space-model","type":"notEquivalentTo"},{"target":"chunking","type":"related"},{"target":"cross-modal-retrieval","type":"related"},{"target":"mteb","type":"related"},{"target":"reverse-image-search","type":"related"}],"distinctions":[{"reason":"A dense encoder is a component or model that produces dense representations; an embedding is the produced representation.","statement":"Embedding is not Dense Encoder.","target":"dense-encoder","targetLabel":"Dense Encoder","targetUri":"https://id.searchplex.net/dense-encoder/"},{"reason":"An embedding is a representation; dense retrieval is a retrieval approach that may use dense embeddings.","statement":"Embedding is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Vector search retrieves over vector representations; an embedding is one such representation.","statement":"Embedding is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"embedding","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Google Machine Learning Crash Course: Embeddings","type":"officialHomepage","url":"https://developers.google.com/machine-learning/crash-course/embeddings"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Embedding","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"A dense encoder is a component or model that produces dense representations; an embedding is the produced representation.","target":"dense-encoder","type":"notEquivalentTo"},{"note":"An embedding is a representation; dense retrieval is a retrieval approach that may use dense embeddings.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Vector search retrieves over vector representations; an embedding is one such representation.","target":"vector-search","type":"notEquivalentTo"},{"target":"dense-encoder","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"multi-vector-retrieval","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/embedding/","json":"https://id.searchplex.net/embedding.json","jsonld":"https://id.searchplex.net/embedding.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In retrieval systems, embeddings are commonly used as vector representations for similarity search, dense retrieval, reranking, clustering, or generation support. An embedding is the representation object, not the model that produced it and not the retrieval system that searches over it.","status":"published","uri":"https://id.searchplex.net/embedding/"}
{"altLabels":["Entity Disambiguation"],"definition":"Mapping an entity mention in a query, document, or other text to a canonical entity in a knowledge base, catalog, graph, or other authority.","distinctions":[{"reason":"Entity linking resolves mentions to entities; entity search retrieves entities or entity-related information.","statement":"Entity Linking is not Entity Search.","target":"entity-search","targetLabel":"Entity Search","targetUri":"https://id.searchplex.net/entity-search/"},{"reason":"Named entity recognition identifies mentions and types; entity linking maps a mention to a canonical entity.","statement":"Entity Linking is not Named Entity Recognition.","target":"named-entity-recognition","targetLabel":"Named Entity Recognition","targetUri":"https://id.searchplex.net/named-entity-recognition/"}],"id":"entity-linking","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Entity Linking and Named Entity Disambiguation","type":"canonicalSurvey","url":"https://doi.org/10.2200/S00494ED1V01Y201305HLT020"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Entity Linking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Entity linking resolves mentions to entities; entity search retrieves entities or entity-related information.","target":"entity-search","type":"notEquivalentTo"},{"note":"Named entity recognition identifies mentions and types; entity linking maps a mention to a canonical entity.","target":"named-entity-recognition","type":"notEquivalentTo"},{"note":"Often used in query understanding when recognized query mentions must be mapped to catalog, graph, or knowledge-base entities.","target":"query-understanding","type":"partOf"},{"target":"entity-search","type":"related"},{"target":"named-entity-recognition","type":"related"}],"representations":{"html":"https://id.searchplex.net/entity-linking/","json":"https://id.searchplex.net/entity-linking.json","jsonld":"https://id.searchplex.net/entity-linking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Entity linking is the canonical-resolution step. Named entity recognition identifies and types mentions; entity linking resolves those mentions to specific entities.","status":"published","terms":[{"community":"general","label":"Entity Resolution","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/entity-linking/"}
{"altLabels":["Entity-Oriented Search","Entity Retrieval"],"definition":"A retrieval task focused on finding entities, entity lists, or entity-related information that satisfies an information need.","derivedRelations":[{"note":"Document retrieval targets documents; entity search targets entities or entity-oriented information.","target":"document-retrieval","type":"notEquivalentTo"}],"distinctions":[{"reason":"Entity search targets entities or entity attributes; ad hoc retrieval is typically framed as retrieving relevant documents or passages for a topic.","statement":"Entity Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Entity search retrieves entities or entity-related information; entity linking maps mentions to canonical entities.","statement":"Entity Search is not Entity Linking.","target":"entity-linking","targetLabel":"Entity Linking","targetUri":"https://id.searchplex.net/entity-linking/"},{"reason":"Semantic search is a broad meaning-oriented search goal; entity search is a task centered on entities.","statement":"Entity Search is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"}],"id":"entity-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC Entity Track","type":"officialHomepage","url":"https://trec.nist.gov/data/entity.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Entity Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Entity search targets entities or entity attributes; ad hoc retrieval is typically framed as retrieving relevant documents or passages for a topic.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Entity search retrieves entities or entity-related information; entity linking maps mentions to canonical entities.","target":"entity-linking","type":"notEquivalentTo"},{"note":"Semantic search is a broad meaning-oriented search goal; entity search is a task centered on entities.","target":"semantic-search","type":"notEquivalentTo"},{"target":"entity-linking","type":"related"},{"target":"semantic-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/entity-search/","json":"https://id.searchplex.net/entity-search.json","jsonld":"https://id.searchplex.net/entity-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Entity search is broader than retrieving documents that mention an entity. TREC Entity Track tasks included entity-oriented search on Web data, such as finding entities and properties of entities.","status":"published","uri":"https://id.searchplex.net/entity-search/"}
{"altLabels":["Exact-Match Search","Exact Phrase Search","Phrase Search"],"definition":"A matching method or query constraint that requires specified terms, values, or phrases to appear exactly as requested.","derivedRelations":[{"note":"Exact match is a matching constraint; Boolean retrieval is a query model using Boolean operators over terms or conditions.","target":"boolean-retrieval","type":"notEquivalentTo"},{"note":"Known-item search describes a user goal of finding a specific item; exact match is a matching method that may help when the user knows an exact title, identifier, or phrase.","target":"known-item-search","type":"notEquivalentTo"},{"note":"Exact match depends on precise terms, values, or phrases; tip-of-the-tongue search is needed when the user lacks the exact identifier.","target":"tip-of-the-tongue-search","type":"notEquivalentTo"},{"target":"boolean-retrieval","type":"related"},{"target":"known-item-search","type":"related"}],"distinctions":[{"reason":"BM25 scores documents using term statistics; exact match requires specified terms, values, or phrases to match exactly.","statement":"Exact Match is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"Exact match is a specific matching constraint; keyword search is a broader search style that may or may not require exact matching.","statement":"Exact Match is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"Exact match is one lexical matching behavior; lexical retrieval is the broader retrieval family using lexical features.","statement":"Exact Match is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"},{"reason":"Exact match requires literal matching; query relaxation loosens matching requirements.","statement":"Exact Match is not Query Relaxation.","target":"query-relaxation","targetLabel":"Query Relaxation","targetUri":"https://id.searchplex.net/query-relaxation/"},{"reason":"Exact match requires literal matching; semantic search is meaning-oriented and may retrieve results without exact lexical overlap.","statement":"Exact Match is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"}],"id":"exact-match","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Phrases and proximity","type":"canonicalSurvey","url":"https://www.dcc.ufmg.br/irbook/10/node25.html"},{"label":"Meilisearch phrase search","type":"officialHomepage","url":"https://meilisearch.com/docs/capabilities/full_text_search/getting_started/phrase_search"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Exact Match","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BM25 scores documents using term statistics; exact match requires specified terms, values, or phrases to match exactly.","target":"bm25","type":"notEquivalentTo"},{"note":"Exact match is a specific matching constraint; keyword search is a broader search style that may or may not require exact matching.","target":"keyword-search","type":"notEquivalentTo"},{"note":"Exact match is one lexical matching behavior; lexical retrieval is the broader retrieval family using lexical features.","target":"lexical-retrieval","type":"notEquivalentTo"},{"note":"Exact match requires literal matching; query relaxation loosens matching requirements.","target":"query-relaxation","type":"notEquivalentTo"},{"note":"Exact match requires literal matching; semantic search is meaning-oriented and may retrieve results without exact lexical overlap.","target":"semantic-search","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"inverted-index","type":"related"},{"target":"keyword-search","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"query-relaxation","type":"related"}],"representations":{"html":"https://id.searchplex.net/exact-match/","json":"https://id.searchplex.net/exact-match.json","jsonld":"https://id.searchplex.net/exact-match.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Exact match may refer to exact field-value matching, exact term matching, or exact phrase matching depending on the system. It is narrower than lexical retrieval generally and should not be used as a synonym for BM25 or keyword search.","status":"published","terms":[{"community":"industry","label":"Quoted Search","usage":"informal"},{"community":"industry","label":"Phrase Match","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/exact-match/"}
{"altLabels":["Faceted Navigation","Faceted Browsing"],"definition":"A search and navigation method that lets users narrow or explore a result set through multiple structured dimensions, called facets.","derivedRelations":[{"note":"Faceted search is an interactive refinement pattern; information filtering is a task formulation for selecting relevant incoming information.","target":"information-filtering","type":"notEquivalentTo"}],"distinctions":[{"reason":"Faceted search exposes structured dimensions for navigation and refinement; result filtering applies constraints to include or exclude results.","statement":"Faceted Search is not Result Filtering.","target":"result-filtering","targetLabel":"Result Filtering","targetUri":"https://id.searchplex.net/result-filtering/"}],"id":"faceted-search","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Faceted metadata for image search and browsing","type":"definingReference","url":"https://dl.acm.org/doi/10.1145/642611.642681"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Faceted Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Faceted search exposes structured dimensions for navigation and refinement; result filtering applies constraints to include or exclude results.","target":"result-filtering","type":"notEquivalentTo"},{"target":"interactive-search","type":"related"},{"target":"result-filtering","type":"related"},{"target":"serp","type":"related"}],"representations":{"html":"https://id.searchplex.net/faceted-search/","json":"https://id.searchplex.net/faceted-search.json","jsonld":"https://id.searchplex.net/faceted-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Faceted search, also called faceted navigation or faceted browsing, uses facets such as author, topic, brand, language, date, or format to expose available refinements. It commonly applies filters, but the concept includes the faceted organization and interactive navigation model rather than filtering alone.","status":"published","terms":[{"community":"industry","label":"Faceting","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/faceted-search/"}
{"altLabels":["Federated Information Retrieval","Distributed Information Retrieval"],"definition":"A retrieval task or system pattern in which a query is sent to multiple search services, collections, indexes, or databases and their results are combined for the user.","derivedRelations":[{"note":"Query routing may be used inside federated search for source selection, but federated search is the broader multi-source retrieval pattern.","target":"query-routing","type":"notEquivalentTo"}],"distinctions":[{"reason":"Federated search searches multiple sources and merges results; ad hoc retrieval is typically evaluated against one known fixed collection.","statement":"Federated Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Result fusion is a combining operation; federated search is the broader multi-source search task or system pattern that often uses result fusion.","statement":"Federated Search is not Result Fusion.","target":"result-fusion","targetLabel":"Result Fusion","targetUri":"https://id.searchplex.net/result-fusion/"}],"id":"federated-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Federated Search","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000010"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Federated Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Federated search searches multiple sources and merges results; ad hoc retrieval is typically evaluated against one known fixed collection.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Result fusion is a combining operation; federated search is the broader multi-source search task or system pattern that often uses result fusion.","target":"result-fusion","type":"notEquivalentTo"},{"target":"query-routing","type":"related"},{"target":"ranking","type":"related"},{"target":"result-fusion","type":"related"}],"representations":{"html":"https://id.searchplex.net/federated-search/","json":"https://id.searchplex.net/federated-search.json","jsonld":"https://id.searchplex.net/federated-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Federated search commonly involves source selection, query dispatch or translation, result merging, and presentation of a unified result set. It is distinct from ad hoc retrieval over a single known collection.","status":"published","uri":"https://id.searchplex.net/federated-search/"}
{"altLabels":["Fact Extraction and VERification"],"artifactKind":"Dataset","definition":"A fact-verification dataset used for claim verification and evidence retrieval evaluation.","distinctions":[{"reason":"FEVER is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"FEVER is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"}],"id":"fever","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"FEVER","type":"officialHomepage","url":"https://fever.ai/"},{"label":"FEVER in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/fever"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"FEVER","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"FEVER is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/fever/","json":"https://id.searchplex.net/fever.json","jsonld":"https://id.searchplex.net/fever.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"FEVER is included in BEIR as a fact-checking retrieval dataset. It is a dataset/test collection, not the BEIR benchmark suite.","status":"published","uri":"https://id.searchplex.net/fever/"}
{"altLabels":["FiQA"],"artifactKind":"Dataset","definition":"A financial question-answering and retrieval evaluation dataset used for information-retrieval benchmarking.","distinctions":[{"reason":"FiQA-2018 is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"FiQA-2018 is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"FiQA-2018 is a dataset used for financial QA/retrieval evaluation; question answering is the task.","statement":"FiQA-2018 is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"}],"id":"fiqa-2018","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"FiQA","type":"officialHomepage","url":"https://sites.google.com/view/fiqa/"},{"label":"FiQA in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/fiqa"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"FiQA-2018","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"FiQA-2018 is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"note":"FiQA-2018 is a dataset used for financial QA/retrieval evaluation; question answering is the task.","target":"question-answering","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"qrels","type":"related"},{"target":"question-answering","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/fiqa-2018/","json":"https://id.searchplex.net/fiqa-2018.json","jsonld":"https://id.searchplex.net/fiqa-2018.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"FiQA-2018 appears in BEIR as a financial-domain retrieval dataset. It is an evaluation dataset, not the broader BEIR benchmark suite.","status":"published","uri":"https://id.searchplex.net/fiqa-2018/"}
{"altLabels":["Forum for Information Retrieval Evaluation"],"definition":"The Forum for Information Retrieval Evaluation, a South Asian evaluation program and meeting for information-retrieval and information-access systems.","distinctions":[{"reason":"FIRE and CLEF are distinct evaluation programs and communities, even though both support multilingual information-access evaluation.","statement":"FIRE is not CLEF.","target":"clef","targetLabel":"CLEF","targetUri":"https://id.searchplex.net/clef/"},{"reason":"FIRE is an evaluation program and meeting; a test collection is an evaluation resource produced or used by specific FIRE tasks.","statement":"FIRE is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"},{"reason":"FIRE and TREC are distinct evaluation programs, though FIRE was modeled as a regional counterpart to programs such as TREC, CLEF, and NTCIR.","statement":"FIRE is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"fire","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"FIRE","type":"officialHomepage","url":"https://fire.irsi.org.in/"},{"label":"FIRE: Forum for Information Retrieval Evaluation","type":"definingReference","url":"https://aclanthology.org/W08-0804/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"FIRE","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"FIRE and CLEF are distinct evaluation programs and communities, even though both support multilingual information-access evaluation.","target":"clef","type":"notEquivalentTo"},{"note":"FIRE is an evaluation program and meeting; a test collection is an evaluation resource produced or used by specific FIRE tasks.","target":"test-collection","type":"notEquivalentTo"},{"note":"FIRE and TREC are distinct evaluation programs, though FIRE was modeled as a regional counterpart to programs such as TREC, CLEF, and NTCIR.","target":"trec","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"cross-language-retrieval","type":"related"},{"target":"multilingual-retrieval","type":"related"},{"target":"ntcir","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/fire/","json":"https://id.searchplex.net/fire.json","jsonld":"https://id.searchplex.net/fire.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"FIRE was started in 2008 and is commonly described as a South Asian counterpart to TREC, CLEF, and NTCIR. It has emphasized multilingual information access, Indian languages, mixed-script retrieval, legal information access, and related evaluation tasks.","status":"published","uri":"https://id.searchplex.net/fire/"}
{"altLabels":["Initial Ranking"],"definition":"Producing an initial ranked ordering of items before later-stage reranking or fusion.","derivedRelations":[{"target":"dual-encoder","type":"uses"}],"distinctions":[{"reason":"First-stage ranking orders items; candidate retrieval may only select a candidate set.","statement":"First-Stage Ranking is not Candidate Retrieval.","target":"candidate-retrieval","targetLabel":"Candidate Retrieval","targetUri":"https://id.searchplex.net/candidate-retrieval/"}],"id":"first-stage-ranking","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Retrieve & Re-Rank","type":"officialHomepage","url":"https://www.sbert.net/examples/applications/retrieve_rerank/README.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"First-Stage Ranking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"ranking","type":"broader"},{"note":"First-stage ranking orders items; candidate retrieval may only select a candidate set.","target":"candidate-retrieval","type":"notEquivalentTo"},{"target":"multi-stage-ranking","type":"partOf"},{"target":"candidate-retrieval","type":"related"},{"target":"reranking","type":"related"}],"representations":{"html":"https://id.searchplex.net/first-stage-ranking/","json":"https://id.searchplex.net/first-stage-ranking.json","jsonld":"https://id.searchplex.net/first-stage-ranking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Use this when the first stage produces an ordered list, not merely an unordered candidate set.","status":"published","terms":[{"community":"industry","label":"Fast Ranking","usage":"informal"}],"uri":"https://id.searchplex.net/first-stage-ranking/"}
{"altLabels":["Graded Relevance Metric","Graded Metric"],"definition":"A family of evaluation metrics designed to use multi-level relevance judgments rather than only binary relevant/not-relevant labels.","derivedRelations":[{"target":"ndcg","type":"narrower"}],"id":"graded-relevance-metric","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Cumulated gain-based evaluation paper","type":"definingReference","url":"https://doi.org/10.1145/582415.582418"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Graded-Relevance Metric","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"qrels","type":"related"},{"target":"rank-sensitive-metric","type":"related"},{"target":"relevance-judgment","type":"related"}],"representations":{"html":"https://id.searchplex.net/graded-relevance-metric/","json":"https://id.searchplex.net/graded-relevance-metric.json","jsonld":"https://id.searchplex.net/graded-relevance-metric.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Graded-relevance metrics are useful when some results are more relevant than others. nDCG is the common example in IR evaluation.","status":"published","uri":"https://id.searchplex.net/graded-relevance-metric/"}
{"altLabels":["AI Grounding"],"definition":"The practice or goal of anchoring generated or system output in supplied, retrieved, or otherwise authoritative source information.","derivedRelations":[{"note":"Question answering is a task objective; grounding concerns whether an answer or generated response is supported by source material.","target":"question-answering","type":"notEquivalentTo"},{"note":"Semantic search retrieves by meaning; grounding concerns whether output is supported by source material.","target":"semantic-search","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"question-answering","type":"related"}],"distinctions":[{"reason":"RAG is one architecture pattern that can support grounding; grounding is the broader output-quality goal of tying claims to source material.","statement":"Grounding is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"},{"reason":"Retrieval selects source material; grounding concerns whether output is supported by source material.","statement":"Grounding is not Retrieval.","target":"retrieval","targetLabel":"Retrieval","targetUri":"https://id.searchplex.net/retrieval/"}],"id":"grounding","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Google Cloud: RAG and grounding","type":"officialHomepage","url":"https://cloud.google.com/use-cases/retrieval-augmented-generation"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Grounding","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"RAG is one architecture pattern that can support grounding; grounding is the broader output-quality goal of tying claims to source material.","target":"rag","type":"notEquivalentTo"},{"note":"Retrieval selects source material; grounding concerns whether output is supported by source material.","target":"retrieval","type":"notEquivalentTo"},{"target":"rag","type":"related"},{"target":"retrieval","type":"related"},{"target":"semantic-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/grounding/","json":"https://id.searchplex.net/grounding.json","jsonld":"https://id.searchplex.net/grounding.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In AI and RAG systems, grounding is often discussed as a way to reduce unsupported or fabricated claims. Retrieval can support grounding, but grounding may also use provided context, tools, databases, citations, or other evidence sources.","status":"published","uri":"https://id.searchplex.net/grounding/"}
{"altLabels":["Hierarchical Navigable Small World"],"definition":"A graph-based approximate nearest neighbor search method using hierarchical navigable small-world graphs.","distinctions":[{"reason":"HNSW is one graph-based approximate nearest neighbor method; ANN search is the broader method family.","statement":"HNSW is not Approximate Nearest Neighbor Search.","target":"approximate-nearest-neighbor-search","targetLabel":"Approximate Nearest Neighbor Search","targetUri":"https://id.searchplex.net/approximate-nearest-neighbor-search/"},{"reason":"HNSW can serve dense retrieval systems, but dense retrieval is the retrieval approach using dense representations.","statement":"HNSW is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"HNSW is an indexing and search method often used for vector search; vector search is the broader retrieval mechanism and product term.","statement":"HNSW is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"hnsw","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Efficient and robust approximate nearest neighbor search using HNSW","type":"definingReference","url":"https://arxiv.org/abs/1603.09320"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"HNSW","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"HNSW is one graph-based approximate nearest neighbor method; ANN search is the broader method family.","target":"approximate-nearest-neighbor-search","type":"notEquivalentTo"},{"note":"HNSW can serve dense retrieval systems, but dense retrieval is the retrieval approach using dense representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"HNSW is an indexing and search method often used for vector search; vector search is the broader retrieval mechanism and product term.","target":"vector-search","type":"notEquivalentTo"},{"target":"approximate-nearest-neighbor-search","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"vector-search","type":"related"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/hnsw/","json":"https://id.searchplex.net/hnsw.json","jsonld":"https://id.searchplex.net/hnsw.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"HNSW is commonly used as a vector-search index method in production systems. It is one ANN method, not a synonym for vector search, dense retrieval, or nearest neighbor search generally.","status":"published","uri":"https://id.searchplex.net/hnsw/"}
{"artifactKind":"Dataset","definition":"A multi-hop question-answering dataset used in retrieval and open-domain QA evaluation.","distinctions":[{"reason":"HotpotQA is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"HotpotQA is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"HotpotQA is a dataset; question answering is the task.","statement":"HotpotQA is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"}],"id":"hotpotqa","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"HotpotQA","type":"officialHomepage","url":"https://hotpotqa.github.io/"},{"label":"HotpotQA in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/hotpotqa"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"HotpotQA","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"HotpotQA is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"note":"HotpotQA is a dataset; question answering is the task.","target":"question-answering","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"question-answering","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/hotpotqa/","json":"https://id.searchplex.net/hotpotqa.json","jsonld":"https://id.searchplex.net/hotpotqa.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"HotpotQA appears in BEIR as a question-answering retrieval dataset. It is a dataset, not the broader question-answering task or the BEIR benchmark suite.","status":"published","uri":"https://id.searchplex.net/hotpotqa/"}
{"altLabels":["Hybrid Search"],"definition":"Retrieval that combines evidence, candidates, or scores from multiple retrieval approaches or representations.","derivedRelations":[{"target":"rag","type":"related"},{"target":"semantic-search","type":"related"}],"distinctions":[{"reason":"Hybrid retrieval commonly combines dense retrieval with lexical or sparse retrieval; it is not the same as dense retrieval alone.","statement":"Hybrid Retrieval is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Reciprocal Rank Fusion is one rank-fusion method that may combine hybrid retrieval results; it is not the hybrid retrieval concept itself.","statement":"Hybrid Retrieval is not Reciprocal Rank Fusion.","target":"reciprocal-rank-fusion","targetLabel":"Reciprocal Rank Fusion","targetUri":"https://id.searchplex.net/reciprocal-rank-fusion/"},{"reason":"Score normalization may be used to combine hybrid retrieval signals, but hybrid retrieval is the broader pattern of combining retrieval approaches.","statement":"Hybrid Retrieval is not Score Normalization.","target":"score-normalization","targetLabel":"Score Normalization","targetUri":"https://id.searchplex.net/score-normalization/"},{"reason":"Hybrid retrieval often includes dense single-vector retrieval, but hybrid retrieval names a combination of retrieval signals rather than the single-vector representation format itself.","statement":"Hybrid Retrieval is not Single-Vector Retrieval.","target":"single-vector-retrieval","targetLabel":"Single-Vector Retrieval","targetUri":"https://id.searchplex.net/single-vector-retrieval/"}],"id":"hybrid-retrieval","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Pinecone hybrid search overview","type":"officialHomepage","url":"https://docs.pinecone.io/guides/search/hybrid-search"},{"label":"Weaviate hybrid search overview","type":"officialHomepage","url":"https://weaviate.io/hybrid-search"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Hybrid Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Hybrid retrieval commonly combines dense retrieval with lexical or sparse retrieval; it is not the same as dense retrieval alone.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Reciprocal Rank Fusion is one rank-fusion method that may combine hybrid retrieval results; it is not the hybrid retrieval concept itself.","target":"reciprocal-rank-fusion","type":"notEquivalentTo"},{"note":"Score normalization may be used to combine hybrid retrieval signals, but hybrid retrieval is the broader pattern of combining retrieval approaches.","target":"score-normalization","type":"notEquivalentTo"},{"note":"Hybrid retrieval often includes dense single-vector retrieval, but hybrid retrieval names a combination of retrieval signals rather than the single-vector representation format itself.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"keyword-search","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"rank-fusion","type":"related"},{"target":"reciprocal-rank-fusion","type":"related"},{"target":"result-fusion","type":"related"},{"target":"score-fusion","type":"related"},{"target":"score-normalization","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"sparse-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/hybrid-retrieval/","json":"https://id.searchplex.net/hybrid-retrieval.json","jsonld":"https://id.searchplex.net/hybrid-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In current production search and RAG usage, hybrid retrieval most often means combining BM25 or lexical retrieval with dense vector retrieval, commonly single-vector retrieval. The concept is broader than that default pairing and does not imply one particular fusion algorithm.","status":"published","uri":"https://id.searchplex.net/hybrid-retrieval/"}
{"altLabels":["Image Search","Visual Search"],"definition":"A retrieval task in which the target items are images or image-like visual objects.","derivedRelations":[{"target":"reverse-image-search","type":"narrower"},{"note":"Image retrieval targets images; multimodal retrieval is broader and may involve images, text, video, audio, tables, or combinations of modalities.","target":"multimodal-retrieval","type":"notEquivalentTo"},{"target":"trecvid","type":"related"}],"distinctions":[{"reason":"Image retrieval targets images; cross-modal retrieval crosses between different query and result modalities.","statement":"Image Retrieval is not Cross-Modal Retrieval.","target":"cross-modal-retrieval","targetLabel":"Cross-Modal Retrieval","targetUri":"https://id.searchplex.net/cross-modal-retrieval/"},{"reason":"Image retrieval retrieves images in general; reverse image search specifically uses an image or visual example as the query.","statement":"Image Retrieval is not Reverse Image Search.","target":"reverse-image-search","targetLabel":"Reverse Image Search","targetUri":"https://id.searchplex.net/reverse-image-search/"},{"reason":"Image retrieval names the target modality; semantic search names a meaning-oriented retrieval goal or capability.","statement":"Image Retrieval is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"}],"id":"image-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Content-Based Image Retrieval","type":"canonicalSurvey","url":"https://doi.org/10.1145/3448154"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Image Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"multimodal-retrieval","type":"broader"},{"note":"Image retrieval targets images; cross-modal retrieval crosses between different query and result modalities.","target":"cross-modal-retrieval","type":"notEquivalentTo"},{"note":"Image retrieval retrieves images in general; reverse image search specifically uses an image or visual example as the query.","target":"reverse-image-search","type":"notEquivalentTo"},{"note":"Image retrieval names the target modality; semantic search names a meaning-oriented retrieval goal or capability.","target":"semantic-search","type":"notEquivalentTo"},{"target":"cross-modal-retrieval","type":"related"},{"target":"semantic-search","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/image-retrieval/","json":"https://id.searchplex.net/image-retrieval.json","jsonld":"https://id.searchplex.net/image-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Image retrieval may use text queries, metadata, visual features, embeddings, or example images. In industry it is often called image search or visual search. It is distinct from reverse image search, where the query itself is an image, and from cross-modal retrieval, where query and target modalities differ.","status":"published","terms":[{"community":"commerce-search","label":"Visual Product Search","usage":"nearSynonym"},{"community":"industry","label":"Vision Search","usage":"informal"}],"uri":"https://id.searchplex.net/image-retrieval/"}
{"altLabels":["Initiative for the Evaluation of XML Retrieval","INitiative for the Evaluation of XML Retrieval"],"definition":"The Initiative for the Evaluation of XML Retrieval, an evaluation program for retrieval from structured XML documents.","distinctions":[{"reason":"INEX is a structured-document retrieval evaluation initiative that later ran within CLEF; CLEF is the broader evaluation forum.","statement":"INEX is not CLEF.","target":"clef","targetLabel":"CLEF","targetUri":"https://id.searchplex.net/clef/"},{"reason":"INEX is an evaluation initiative; document retrieval is a retrieval task evaluated by some INEX-style resources.","statement":"INEX is not Document Retrieval.","target":"document-retrieval","targetLabel":"Document Retrieval","targetUri":"https://id.searchplex.net/document-retrieval/"},{"reason":"INEX is an evaluation program; its test collections are evaluation resources used by INEX tasks.","statement":"INEX is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"inex","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of XML retrieval","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-xml-retrieval-1.html"},{"label":"Report on INEX 2012","type":"canonicalSurvey","url":"https://doi.org/10.1145/2505955.2505967"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"INEX","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"INEX is a structured-document retrieval evaluation initiative that later ran within CLEF; CLEF is the broader evaluation forum.","target":"clef","type":"notEquivalentTo"},{"note":"INEX is an evaluation initiative; document retrieval is a retrieval task evaluated by some INEX-style resources.","target":"document-retrieval","type":"notEquivalentTo"},{"note":"INEX is an evaluation program; its test collections are evaluation resources used by INEX tasks.","target":"test-collection","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/inex/","json":"https://id.searchplex.net/inex.json","jsonld":"https://id.searchplex.net/inex.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"INEX focused on structured-document and focused retrieval, including evaluation of systems that retrieve document elements or passages rather than only whole documents. Since 2012, INEX has been run as part of CLEF.","status":"published","uri":"https://id.searchplex.net/inex/"}
{"altLabels":["Filtering"],"definition":"An IR task in which a system identifies items from a stream or changing collection that satisfy a relatively stable information need, profile, or topic.","distinctions":[{"reason":"Information filtering uses stable needs or profiles over incoming items; ad hoc retrieval uses new queries or topics against a known collection.","statement":"Information Filtering is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Faceted search is an interactive refinement pattern; information filtering is a task formulation for selecting relevant incoming information.","statement":"Information Filtering is not Faceted Search.","target":"faceted-search","targetLabel":"Faceted Search","targetUri":"https://id.searchplex.net/faceted-search/"},{"reason":"Information filtering selects relevant incoming items for a standing need; result filtering applies constraints to a result set or retrievable subset.","statement":"Information Filtering is not Result Filtering.","target":"result-filtering","targetLabel":"Result Filtering","targetUri":"https://id.searchplex.net/result-filtering/"}],"id":"information-filtering","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"The TREC-8 Filtering Track Final Report","type":"definingReference","url":"https://trec.nist.gov/pubs/trec8/papers/filtering_new.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Information Filtering","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Information filtering uses stable needs or profiles over incoming items; ad hoc retrieval uses new queries or topics against a known collection.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Faceted search is an interactive refinement pattern; information filtering is a task formulation for selecting relevant incoming information.","target":"faceted-search","type":"notEquivalentTo"},{"note":"Information filtering selects relevant incoming items for a standing need; result filtering applies constraints to a result set or retrievable subset.","target":"result-filtering","type":"notEquivalentTo"},{"target":"relevance-judgment","type":"related"},{"target":"routing","type":"related"}],"representations":{"html":"https://id.searchplex.net/information-filtering/","json":"https://id.searchplex.net/information-filtering.json","jsonld":"https://id.searchplex.net/information-filtering.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Information filtering emphasizes ongoing selection of relevant incoming items, often using persistent user profiles or topics. It is distinct from result filtering, which restricts already retrieved or retrievable results by constraints, and from ad hoc retrieval over a fixed collection.","status":"published","terms":[{"community":"academic","label":"Adaptive Filtering","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/information-filtering/"}
{"altLabels":["Interactive Retrieval","Interactive Information Retrieval"],"definition":"A search task or setting in which users interact with the system over time by reformulating queries, examining results, applying controls, or providing feedback.","derivedRelations":[{"target":"conversational-search","type":"narrower"},{"target":"session-search","type":"narrower"},{"note":"Interactive search emphasizes user-system interaction; iterative retrieval can happen internally without user interaction.","target":"iterative-retrieval","type":"notEquivalentTo"},{"note":"Known-item search describes the user's target; interactive search describes an interaction style that may include known-item, exploratory, or other goals.","target":"known-item-search","type":"notEquivalentTo"},{"note":"Single-shot retrieval produces results without user-system interaction over time; interactive search centers on interaction across a search process.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"known-item-search","type":"related"},{"target":"tip-of-the-tongue-search","type":"related"}],"distinctions":[{"reason":"Interactive search centers on user-system interaction across a session; ad hoc retrieval is often evaluated as a single-shot retrieval task.","statement":"Interactive Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Conversational search is a dialogue-oriented form of interactive search; interactive search also includes non-conversational interactions such as faceting, browsing, and reformulation.","statement":"Interactive Search is not Conversational Search.","target":"conversational-search","targetLabel":"Conversational Search","targetUri":"https://id.searchplex.net/conversational-search/"}],"id":"interactive-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Interactive Information Retrieval","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000023"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Interactive Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Interactive search centers on user-system interaction across a session; ad hoc retrieval is often evaluated as a single-shot retrieval task.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Conversational search is a dialogue-oriented form of interactive search; interactive search also includes non-conversational interactions such as faceting, browsing, and reformulation.","target":"conversational-search","type":"notEquivalentTo"},{"target":"faceted-search","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"serp","type":"related"},{"target":"single-shot-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/interactive-search/","json":"https://id.searchplex.net/interactive-search.json","jsonld":"https://id.searchplex.net/interactive-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Interactive search emphasizes the search session and user-system interaction rather than a single batch retrieval run. It may involve browsing, faceting, relevance feedback, query reformulation, and exploratory behavior. It is not the same as iterative retrieval: interaction is about user-system exchange, while iteration is about repeated retrieval attempts.","status":"published","uri":"https://id.searchplex.net/interactive-search/"}
{"definition":"An index structure that maps terms or other lexical features to postings lists of documents or items in which those features occur.","derivedRelations":[{"target":"boolean-retrieval","type":"related"},{"target":"exact-match","type":"related"}],"distinctions":[{"reason":"An inverted index provides access to postings; BM25 is a scoring method that can use those postings.","statement":"Inverted Index is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"Keyword search is a user-facing search style; an inverted index is one common implementation structure beneath lexical search.","statement":"Inverted Index is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"An inverted index is an index structure; lexical retrieval is the broader retrieval approach that may use inverted indexes and term-based scoring.","statement":"Inverted Index is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"}],"id":"inverted-index","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"A first take at building an inverted index","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/a-first-take-at-building-an-inverted-index-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Inverted Index","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"An inverted index provides access to postings; BM25 is a scoring method that can use those postings.","target":"bm25","type":"notEquivalentTo"},{"note":"Keyword search is a user-facing search style; an inverted index is one common implementation structure beneath lexical search.","target":"keyword-search","type":"notEquivalentTo"},{"note":"An inverted index is an index structure; lexical retrieval is the broader retrieval approach that may use inverted indexes and term-based scoring.","target":"lexical-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"keyword-search","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/inverted-index/","json":"https://id.searchplex.net/inverted-index.json","jsonld":"https://id.searchplex.net/inverted-index.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Inverted indexes are foundational infrastructure for lexical retrieval and keyword search. They enable efficient lookup of lexical evidence, but they are not the retrieval task, a ranking function, or a search engine by themselves.","status":"published","uri":"https://id.searchplex.net/inverted-index/"}
{"altLabels":["ir-datasets"],"artifactKind":"Software","definition":"A Python package and catalog that provides a common interface for accessing, documenting, and working with many information-retrieval datasets and benchmarks.","distinctions":[{"reason":"ir_datasets provides access to BEIR and many other datasets; BEIR is a benchmark suite.","statement":"ir_datasets is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"ir_datasets can expose qrels; qrels are relevance-judgment files, not the software package.","statement":"ir_datasets is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"ir_datasets is software and catalog infrastructure for working with test collections; a test collection is an evaluation resource.","statement":"ir_datasets is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"ir-datasets","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ir_datasets catalog","type":"officialHomepage","url":"https://ir-datasets.com/"},{"label":"ir_datasets repository","type":"officialRepository","url":"https://github.com/allenai/ir_datasets"},{"label":"Simplified Data Wrangling with ir_datasets","type":"definingReference","url":"https://arxiv.org/abs/2103.02280"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"ir_datasets","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"ir_datasets provides access to BEIR and many other datasets; BEIR is a benchmark suite.","target":"beir","type":"notEquivalentTo"},{"note":"ir_datasets can expose qrels; qrels are relevance-judgment files, not the software package.","target":"qrels","type":"notEquivalentTo"},{"note":"ir_datasets is software and catalog infrastructure for working with test collections; a test collection is an evaluation resource.","target":"test-collection","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"clef","type":"related"},{"target":"corpus","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"nfcorpus","type":"related"},{"target":"qrels","type":"related"},{"target":"scifact","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/ir-datasets/","json":"https://id.searchplex.net/ir-datasets.json","jsonld":"https://id.searchplex.net/ir-datasets.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"ir_datasets helps acquire and iterate over IR dataset components such as documents, queries, and relevance judgments. It is not itself a dataset, benchmark suite, qrels file, or evaluation metric.","status":"published","terms":[{"community":"academic","label":"IR datasets catalog","usage":"nearSynonym"},{"community":"industry","label":"IR dataset loader","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/ir-datasets/"}
{"altLabels":["Iterative Search"],"definition":"A retrieval pattern in which retrieval is performed repeatedly, with later retrieval attempts informed by earlier results, feedback, reasoning, reformulation, or intermediate state.","derivedRelations":[{"target":"multi-hop-retrieval","type":"narrower"}],"distinctions":[{"reason":"Iterative retrieval may be agent-driven, user-driven, or system-driven; agentic retrieval specifically involves an agent controlling the retrieval loop.","statement":"Iterative Retrieval is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Interactive search emphasizes user-system interaction; iterative retrieval can happen internally without user interaction.","statement":"Iterative Retrieval is not Interactive Search.","target":"interactive-search","targetLabel":"Interactive Search","targetUri":"https://id.searchplex.net/interactive-search/"},{"reason":"Iterative retrieval repeats retrieval attempts; multi-hop retrieval specifically uses multiple dependent evidence hops or subproblems.","statement":"Iterative Retrieval is not Multi-Hop Retrieval.","target":"multi-hop-retrieval","targetLabel":"Multi-Hop Retrieval","targetUri":"https://id.searchplex.net/multi-hop-retrieval/"},{"reason":"Query reformulation changes a query; iterative retrieval is the broader repeated retrieval pattern that may use reformulation between attempts.","statement":"Iterative Retrieval is not Query Reformulation.","target":"query-reformulation","targetLabel":"Query Reformulation","targetUri":"https://id.searchplex.net/query-reformulation/"},{"reason":"Iterative retrieval performs repeated retrieval attempts; single-shot retrieval performs one non-iterative retrieval attempt.","statement":"Iterative Retrieval is not Single-Shot Retrieval.","target":"single-shot-retrieval","targetLabel":"Single-Shot Retrieval","targetUri":"https://id.searchplex.net/single-shot-retrieval/"}],"id":"iterative-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Relevance feedback and query expansion","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html"},{"label":"Generative Multi-hop Retrieval","type":"definingReference","url":"https://arxiv.org/abs/2204.13596"},{"label":"Agentic Retrieval-Augmented Generation","type":"canonicalSurvey","url":"https://arxiv.org/abs/2501.09136"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Iterative Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Iterative retrieval may be agent-driven, user-driven, or system-driven; agentic retrieval specifically involves an agent controlling the retrieval loop.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Interactive search emphasizes user-system interaction; iterative retrieval can happen internally without user interaction.","target":"interactive-search","type":"notEquivalentTo"},{"note":"Iterative retrieval repeats retrieval attempts; multi-hop retrieval specifically uses multiple dependent evidence hops or subproblems.","target":"multi-hop-retrieval","type":"notEquivalentTo"},{"note":"Query reformulation changes a query; iterative retrieval is the broader repeated retrieval pattern that may use reformulation between attempts.","target":"query-reformulation","type":"notEquivalentTo"},{"note":"Iterative retrieval performs repeated retrieval attempts; single-shot retrieval performs one non-iterative retrieval attempt.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"agentic-retrieval","type":"related"},{"target":"interactive-search","type":"related"},{"target":"pseudo-relevance-feedback","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"retrieval-planning","type":"related"},{"target":"session-search","type":"related"},{"target":"single-shot-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/iterative-retrieval/","json":"https://id.searchplex.net/iterative-retrieval.json","jsonld":"https://id.searchplex.net/iterative-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Iterative retrieval is broader than multi-hop retrieval and broader than agentic retrieval. The iteration may be user-driven, system-driven, feedback-driven, or agent-driven. It is distinct from interactive search, which emphasizes user-system interaction, and from single-shot retrieval, which performs one non-iterative retrieval attempt.","status":"published","terms":[{"community":"industry","label":"Repeated Retrieval","usage":"nearSynonym"},{"community":"industry","label":"Retrieval Loop","usage":"nearSynonym"},{"community":"ml-nlp","label":"Iterative RAG","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/iterative-retrieval/"}
{"altLabels":["Keyword-Based Search","Keyword Retrieval"],"definition":"Search that retrieves results primarily by matching explicit query keywords, terms, or phrases against indexed text or metadata.","derivedRelations":[{"note":"Boolean retrieval uses logical query expressions; keyword search is a broader practitioner-facing search style.","target":"boolean-retrieval","type":"notEquivalentTo"},{"note":"Keyword search is a user-facing search style; an inverted index is one common implementation structure beneath lexical search.","target":"inverted-index","type":"notEquivalentTo"}],"distinctions":[{"reason":"Keyword search is a search style; BM25 is a scoring method that may be used in keyword search.","statement":"Keyword Search is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"Keyword search may support exact-match constraints, but keyword search is broader than exact matching.","statement":"Keyword Search is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"The terms are often used as near-synonyms in industry, but Searchplex ID keeps them separate: keyword search is the user/practitioner-facing search style, while lexical retrieval is the broader technical family based on lexical features.","statement":"Keyword Search is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"},{"reason":"Keyword search relies on explicit lexical evidence from the query; semantic search is meaning-oriented and may use dense, sparse, hybrid, expansion, or other retrieval methods.","statement":"Keyword Search is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Keyword search describes a search style using explicit terms or phrases; sparse retrieval describes a representation property and includes both traditional lexical and learned sparse approaches.","statement":"Keyword Search is not Sparse Retrieval.","target":"sparse-retrieval","targetLabel":"Sparse Retrieval","targetUri":"https://id.searchplex.net/sparse-retrieval/"}],"id":"keyword-search","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Term vocabulary and postings lists","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/term-vocabulary-and-postings-lists-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Keyword Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Keyword search is a search style; BM25 is a scoring method that may be used in keyword search.","target":"bm25","type":"notEquivalentTo"},{"note":"Keyword search may support exact-match constraints, but keyword search is broader than exact matching.","target":"exact-match","type":"notEquivalentTo"},{"note":"The terms are often used as near-synonyms in industry, but Searchplex ID keeps them separate: keyword search is the user/practitioner-facing search style, while lexical retrieval is the broader technical family based on lexical features.","target":"lexical-retrieval","type":"notEquivalentTo"},{"note":"Keyword search relies on explicit lexical evidence from the query; semantic search is meaning-oriented and may use dense, sparse, hybrid, expansion, or other retrieval methods.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Keyword search describes a search style using explicit terms or phrases; sparse retrieval describes a representation property and includes both traditional lexical and learned sparse approaches.","target":"sparse-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"boolean-retrieval","type":"related"},{"target":"exact-match","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"inverted-index","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"semantic-search","type":"related"},{"target":"sparse-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/keyword-search/","json":"https://id.searchplex.net/keyword-search.json","jsonld":"https://id.searchplex.net/keyword-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Keyword search is often used interchangeably with lexical or full-text search in industry. In Searchplex ID, keyword search names the practitioner- and user-facing search style centered on explicit query words or phrases, while lexical retrieval names the broader technical retrieval family based on lexical features.","status":"published","uri":"https://id.searchplex.net/keyword-search/"}
{"altLabels":["Known Item Search","Known-Item Retrieval"],"definition":"A search task in which the user is trying to find a particular known item rather than explore a topic or discover unknown relevant material.","derivedRelations":[{"target":"tip-of-the-tongue-search","type":"narrower"},{"note":"Tip-of-the-tongue search is a specialized form of known-item search where the searcher cannot reliably recall the item identifier.","target":"tip-of-the-tongue-search","type":"notEquivalentTo"}],"distinctions":[{"reason":"Known-item search targets a particular item already in the user's mind; ad hoc retrieval usually targets topical relevance for a new information need.","statement":"Known-Item Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Known-item search describes a user goal of finding a specific item; exact match is a matching method that may help when the user knows an exact title, identifier, or phrase.","statement":"Known-Item Search is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"Known-item search describes the user's target; interactive search describes an interaction style that may include known-item, exploratory, or other goals.","statement":"Known-Item Search is not Interactive Search.","target":"interactive-search","targetLabel":"Interactive Search","targetUri":"https://id.searchplex.net/interactive-search/"}],"id":"known-item-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Combining Document Representations for Known-Item Search","type":"definingReference","url":"https://doi.org/10.1145/860435.860458"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Known-Item Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Known-item search targets a particular item already in the user's mind; ad hoc retrieval usually targets topical relevance for a new information need.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Known-item search describes a user goal of finding a specific item; exact match is a matching method that may help when the user knows an exact title, identifier, or phrase.","target":"exact-match","type":"notEquivalentTo"},{"note":"Known-item search describes the user's target; interactive search describes an interaction style that may include known-item, exploratory, or other goals.","target":"interactive-search","type":"notEquivalentTo"},{"target":"exact-match","type":"related"},{"target":"interactive-search","type":"related"},{"target":"web-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/known-item-search/","json":"https://id.searchplex.net/known-item-search.json","jsonld":"https://id.searchplex.net/known-item-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Known-item search is common in library, web, enterprise, and catalog search. It is distinct from ad hoc topical retrieval and exploratory or interactive search, because the user has a specific target item in mind.","status":"published","terms":[{"community":"industry","label":"Navigational Search","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/known-item-search/"}
{"definition":"A retrieval approach in which queries and documents are independently encoded into fine-grained representations and their interactions are computed at query time rather than fully collapsed into a single vector beforehand.","derivedRelations":[{"target":"colbert","type":"realizedBy"},{"target":"colbertv2","type":"realizedBy"}],"distinctions":[{"reason":"Multi-vector retrieval describes using multiple vectors per item; late interaction describes computing fine-grained interactions at query time.","statement":"Late Interaction is not Multi-Vector Retrieval.","target":"multi-vector-retrieval","targetLabel":"Multi-Vector Retrieval","targetUri":"https://id.searchplex.net/multi-vector-retrieval/"}],"id":"late-interaction","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ColBERT paper","type":"definingReference","url":"https://doi.org/10.1145/3397271.3401075"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Late Interaction","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Multi-vector retrieval describes using multiple vectors per item; late interaction describes computing fine-grained interactions at query time.","target":"multi-vector-retrieval","type":"notEquivalentTo"},{"target":"maxsim","type":"related"},{"target":"multi-vector-retrieval","type":"related"},{"target":"plaid","type":"related"}],"representations":{"html":"https://id.searchplex.net/late-interaction/","json":"https://id.searchplex.net/late-interaction.json","jsonld":"https://id.searchplex.net/late-interaction.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"ColBERT popularized token-level late interaction for neural retrieval. Late interaction is a retrieval paradigm rather than a specific model.","status":"published","uri":"https://id.searchplex.net/late-interaction/"}
{"altLabels":["Neural Sparse Retrieval"],"definition":"Retrieval using sparse query and document representations whose feature activations or weights are learned from data.","derivedRelations":[{"note":"Sparse retrieval includes traditional lexical sparse representations as well as learned sparse representations; learned sparse retrieval is the neural learned subset.","target":"sparse-retrieval","type":"notEquivalentTo"},{"target":"splade","type":"realizedBy"},{"target":"dynamic-pruning","type":"related"},{"target":"semantic-search","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"distinctions":[{"reason":"Learned sparse retrieval is a representation family; query expansion is a task or transformation applied to a query.","statement":"Learned Sparse Retrieval is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"},{"reason":"SPLADE is one model family that realizes learned sparse retrieval, not the whole representation family.","statement":"Learned Sparse Retrieval is not SPLADE.","target":"splade","targetLabel":"SPLADE","targetUri":"https://id.searchplex.net/splade/"}],"id":"learned-sparse-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"SPLADE","type":"definingReference","url":"https://doi.org/10.1145/3404835.3463098"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Learned Sparse Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"sparse-retrieval","type":"broader"},{"note":"Learned sparse retrieval is a representation family; query expansion is a task or transformation applied to a query.","target":"query-expansion","type":"notEquivalentTo"},{"note":"SPLADE is one model family that realizes learned sparse retrieval, not the whole representation family.","target":"splade","type":"notEquivalentTo"},{"note":"Learned sparse models may activate expansion-like vocabulary features, but the representation family is not the query-expansion task.","target":"query-expansion","type":"related"}],"representations":{"html":"https://id.searchplex.net/learned-sparse-retrieval/","json":"https://id.searchplex.net/learned-sparse-retrieval.json","jsonld":"https://id.searchplex.net/learned-sparse-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Learned sparse systems commonly retain compatibility with inverted-index-style retrieval while learning lexical or vocabulary-space representations. Expansion-like activations may occur, but the representation family is not itself the query-expansion task.","status":"published","terms":[{"community":"academic","label":"Sparse Neural Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/learned-sparse-retrieval/"}
{"altLabels":["LTR"],"definition":"A family of supervised machine-learning methods for learning ranking functions from training data such as relevance judgments, preferences, clicks, or other labels.","distinctions":[{"reason":"Ranking is the task of ordering results; Learning to Rank is a method family for learning ranking functions.","statement":"Learning to Rank is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"Reranking is a pipeline stage or task; Learning to Rank methods may be used for reranking but are not the stage itself.","statement":"Learning to Rank is not Reranking.","target":"reranking","targetLabel":"Reranking","targetUri":"https://id.searchplex.net/reranking/"}],"id":"learning-to-rank","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Learning to Rank for Information Retrieval","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000016"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Learning to Rank","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Ranking is the task of ordering results; Learning to Rank is a method family for learning ranking functions.","target":"ranking","type":"notEquivalentTo"},{"note":"Reranking is a pipeline stage or task; Learning to Rank methods may be used for reranking but are not the stage itself.","target":"reranking","type":"notEquivalentTo"},{"target":"ndcg","type":"related"},{"target":"position-bias","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"ranking","type":"usedFor"},{"target":"reranking","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/learning-to-rank/","json":"https://id.searchplex.net/learning-to-rank.json","jsonld":"https://id.searchplex.net/learning-to-rank.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Learning to Rank is a method family used to produce or improve rankings. It is not the ranking task itself, and it is not limited to any one model family or stage of a retrieval pipeline.","status":"published","uri":"https://id.searchplex.net/learning-to-rank/"}
{"altLabels":["Lexical Search"],"definition":"Retrieval based primarily on explicit lexical features such as terms, tokens, phrases, or other symbolic features shared between queries and indexed items.","derivedRelations":[{"note":"BM25F is one field-aware lexical scoring method; lexical retrieval is the broader retrieval family.","target":"bm25f","type":"notEquivalentTo"},{"target":"document-expansion","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"distinctions":[{"reason":"Lexical retrieval is the broader retrieval family based on lexical features; BM25 is one lexical scoring method.","statement":"Lexical Retrieval is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"Exact match is a specific lexical matching constraint; lexical retrieval is the broader family of lexical retrieval approaches.","statement":"Lexical Retrieval is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"Lexical retrieval is the retrieval approach; an inverted index is a data structure commonly used to support lexical retrieval.","statement":"Lexical Retrieval is not Inverted Index.","target":"inverted-index","targetLabel":"Inverted Index","targetUri":"https://id.searchplex.net/inverted-index/"},{"reason":"The terms are often used as near-synonyms in industry, but Searchplex ID keeps them separate: lexical retrieval names the broader technical approach based on lexical features, while keyword search names the user/practitioner-facing search style.","statement":"Lexical Retrieval is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"Lexical retrieval is based on explicit lexical features; sparse retrieval is the broader representation pattern that also includes learned sparse models.","statement":"Lexical Retrieval is not Sparse Retrieval.","target":"sparse-retrieval","targetLabel":"Sparse Retrieval","targetUri":"https://id.searchplex.net/sparse-retrieval/"},{"reason":"TF-IDF is a term-weighting method; lexical retrieval is the broader retrieval family based on lexical features.","statement":"Lexical Retrieval is not TF-IDF.","target":"tf-idf","targetLabel":"TF-IDF","targetUri":"https://id.searchplex.net/tf-idf/"}],"id":"lexical-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Term vocabulary and postings lists","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/the-term-vocabulary-and-postings-lists-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Lexical Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Lexical retrieval is the broader retrieval family based on lexical features; BM25 is one lexical scoring method.","target":"bm25","type":"notEquivalentTo"},{"note":"Exact match is a specific lexical matching constraint; lexical retrieval is the broader family of lexical retrieval approaches.","target":"exact-match","type":"notEquivalentTo"},{"note":"Lexical retrieval is the retrieval approach; an inverted index is a data structure commonly used to support lexical retrieval.","target":"inverted-index","type":"notEquivalentTo"},{"note":"The terms are often used as near-synonyms in industry, but Searchplex ID keeps them separate: lexical retrieval names the broader technical approach based on lexical features, while keyword search names the user/practitioner-facing search style.","target":"keyword-search","type":"notEquivalentTo"},{"note":"Lexical retrieval is based on explicit lexical features; sparse retrieval is the broader representation pattern that also includes learned sparse models.","target":"sparse-retrieval","type":"notEquivalentTo"},{"note":"TF-IDF is a term-weighting method; lexical retrieval is the broader retrieval family based on lexical features.","target":"tf-idf","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"bm25f","type":"related"},{"target":"boolean-retrieval","type":"related"},{"target":"exact-match","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"inverted-index","type":"related"},{"target":"keyword-search","type":"related"},{"target":"sparse-retrieval","type":"related"},{"target":"tf-idf","type":"related"},{"target":"vector-space-model","type":"related"}],"representations":{"html":"https://id.searchplex.net/lexical-retrieval/","json":"https://id.searchplex.net/lexical-retrieval.json","jsonld":"https://id.searchplex.net/lexical-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Lexical retrieval commonly uses inverted indexes and term-based scoring, and is often called keyword or full-text search in practitioner material. Searchplex ID treats it as the broader technical retrieval family, not as only the user-facing keyword-search style.","status":"published","uri":"https://id.searchplex.net/lexical-retrieval/"}
{"altLabels":["LLM Judge","LLM-as-a-Judge"],"definition":"A method that uses a large language model to evaluate, label, score, or compare outputs or query-item pairs.","distinctions":[{"reason":"LLM-as-judge is a judging method; qrels are a structured set of relevance labels.","statement":"LLM-as-Judge is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"LLM-as-judge may estimate or label relevance, but it is not the relevance relationship itself.","statement":"LLM-as-Judge is not Relevance.","target":"relevance","targetLabel":"Relevance","targetUri":"https://id.searchplex.net/relevance/"},{"reason":"LLM-as-judge is a method for producing or assisting judgments; a relevance judgment is the assessment itself.","statement":"LLM-as-Judge is not Relevance Judgment.","target":"relevance-judgment","targetLabel":"Relevance Judgment","targetUri":"https://id.searchplex.net/relevance-judgment/"}],"id":"llm-as-judge","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"LLMJudge: LLMs for Relevance Judgments","type":"canonicalSurvey","url":"https://arxiv.org/abs/2408.08896"},{"label":"Don't Use LLMs to Make Relevance Judgments","type":"canonicalSurvey","url":"https://irrj.org/article/id/2564/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"LLM-as-Judge","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"LLM-as-judge is a judging method; qrels are a structured set of relevance labels.","target":"qrels","type":"notEquivalentTo"},{"note":"LLM-as-judge may estimate or label relevance, but it is not the relevance relationship itself.","target":"relevance","type":"notEquivalentTo"},{"note":"LLM-as-judge is a method for producing or assisting judgments; a relevance judgment is the assessment itself.","target":"relevance-judgment","type":"notEquivalentTo"},{"target":"qrels","type":"related"},{"target":"relevance","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/llm-as-judge/","json":"https://id.searchplex.net/llm-as-judge.json","jsonld":"https://id.searchplex.net/llm-as-judge.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In retrieval evaluation, LLM-as-judge methods may be used to generate or assist relevance assessments. They are not the same as human relevance judgments, qrels, or the relevance relationship being judged.","status":"published","uri":"https://id.searchplex.net/llm-as-judge/"}
{"definition":"A late-interaction scoring operation that, for each query representation, selects the maximum similarity to document representations and aggregates those maxima into a relevance score.","derivedRelations":[{"target":"plaid","type":"related"}],"distinctions":[{"reason":"MaxSim is a scoring operation; ColBERT is a model architecture that uses MaxSim-style late interaction.","statement":"MaxSim is not ColBERT.","target":"colbert","targetLabel":"ColBERT","targetUri":"https://id.searchplex.net/colbert/"}],"id":"maxsim","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ColBERT paper","type":"definingReference","url":"https://doi.org/10.1145/3397271.3401075"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"MaxSim","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"MaxSim is a scoring operation; ColBERT is a model architecture that uses MaxSim-style late interaction.","target":"colbert","type":"notEquivalentTo"},{"target":"colbert","type":"related"},{"target":"colbertv2","type":"related"},{"target":"late-interaction","type":"related"},{"target":"ranking","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/maxsim/","json":"https://id.searchplex.net/maxsim.json","jsonld":"https://id.searchplex.net/maxsim.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"MaxSim is a scoring operation used by ColBERT-style late interaction. It is not itself a retrieval model.","status":"published","uri":"https://id.searchplex.net/maxsim/"}
{"altLabels":["MAP","mAP"],"definition":"A ranking evaluation metric that averages Average Precision values across a set of queries or topics.","derivedRelations":[{"target":"mrr","type":"related"}],"distinctions":[{"reason":"MAP averages Average Precision across queries or topics; AP is the per-query or per-topic value.","statement":"Mean Average Precision is not Average Precision.","target":"average-precision","targetLabel":"Average Precision","targetUri":"https://id.searchplex.net/average-precision/"},{"reason":"MAP averages rank-sensitive AP values across topics; Precision@K only measures precision at a fixed cutoff.","statement":"Mean Average Precision is not Precision@K.","target":"precision-at-k","targetLabel":"Precision@K","targetUri":"https://id.searchplex.net/precision-at-k/"}],"id":"mean-average-precision","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Mean Average Precision","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"precision-metric","type":"broader"},{"target":"rank-sensitive-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"MAP averages Average Precision across queries or topics; AP is the per-query or per-topic value.","target":"average-precision","type":"notEquivalentTo"},{"note":"MAP averages rank-sensitive AP values across topics; Precision@K only measures precision at a fixed cutoff.","target":"precision-at-k","type":"notEquivalentTo"},{"target":"average-precision","type":"related"},{"target":"ndcg","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"qrels","type":"related"}],"representations":{"html":"https://id.searchplex.net/mean-average-precision/","json":"https://id.searchplex.net/mean-average-precision.json","jsonld":"https://id.searchplex.net/mean-average-precision.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"MAP summarizes ranked retrieval effectiveness across topics using AP as the per-topic measure. It should not be confused with Average Precision for a single topic or with Precision@K at a fixed cutoff.","status":"published","uri":"https://id.searchplex.net/mean-average-precision/"}
{"altLabels":["Mean Reciprocal Rank"],"definition":"A ranking evaluation metric that averages the reciprocal rank of the first relevant result across queries.","derivedRelations":[{"target":"ms-marco-passage-ranking","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"ranking","type":"related"}],"id":"mrr","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"MRR","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"rank-sensitive-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"reranking","type":"evaluates"},{"target":"mean-average-precision","type":"related"},{"target":"qrels","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"success-at-k","type":"related"}],"representations":{"html":"https://id.searchplex.net/mrr/","json":"https://id.searchplex.net/mrr.json","jsonld":"https://id.searchplex.net/mrr.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"MRR is most appropriate when the position of the first relevant result is the main objective.","status":"published","uri":"https://id.searchplex.net/mrr/"}
{"artifactKind":"Benchmark","definition":"A large-scale passage-retrieval and ranking benchmark derived from real search queries, a corpus of passages, and relevance or answer-based supervision.","id":"ms-marco-passage-ranking","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"MS MARCO","type":"officialHomepage","url":"https://microsoft.github.io/msmarco/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"MS MARCO Passage Ranking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"ir-datasets","type":"related"},{"target":"mrr","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"ranking","type":"related"},{"target":"reranking","type":"related"},{"target":"retrieval","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/ms-marco-passage-ranking/","json":"https://id.searchplex.net/ms-marco-passage-ranking.json","jsonld":"https://id.searchplex.net/ms-marco-passage-ranking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"MS MARCO includes multiple datasets and tasks. This identity refers specifically to the passage-ranking resource or task family and should remain distinct from MS MARCO document ranking and other MS MARCO datasets.","status":"published","terms":[{"community":"academic","label":"MS MARCO Passage","usage":"nearSynonym"},{"community":"academic","label":"MS MARCO Passage Ranking Dataset","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/ms-marco-passage-ranking/"}
{"altLabels":["Massive Text Embedding Benchmark"],"artifactKind":"Benchmark","definition":"The Massive Text Embedding Benchmark, a benchmark suite for evaluating text embedding models across multiple tasks, datasets, and languages.","derivedRelations":[{"note":"BrowseComp-Plus evaluates deep-research agents and retrievers; MTEB evaluates text embedding models across multiple tasks.","target":"browsecomp-plus","type":"notEquivalentTo"},{"note":"MTEB evaluates text embedding models across tasks; web browsing agent evaluation evaluates agent behavior in web information-seeking tasks.","target":"web-browsing-agent-evaluation","type":"notEquivalentTo"}],"distinctions":[{"reason":"MTEB evaluates text embeddings across many NLP and retrieval-adjacent tasks; BEIR is an information-retrieval benchmark suite.","statement":"MTEB is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"MTEB is a benchmark suite for evaluating embedding models; an embedding is a vector representation.","statement":"MTEB is not Embedding.","target":"embedding","targetLabel":"Embedding","targetUri":"https://id.searchplex.net/embedding/"},{"reason":"MTEB is a benchmark suite aggregating many tasks and datasets; a test collection is an evaluation resource pattern.","statement":"MTEB is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"mteb","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"MTEB Leaderboard","type":"officialHomepage","url":"https://huggingface.co/spaces/mteb/leaderboard"},{"label":"MTEB repository","type":"officialRepository","url":"https://github.com/embeddings-benchmark/mteb"},{"label":"MTEB: Massive Text Embedding Benchmark","type":"definingReference","url":"https://arxiv.org/abs/2210.07316"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"MTEB","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"dense-retrieval","type":"evaluates"},{"target":"embedding","type":"evaluates"},{"target":"reranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"MTEB evaluates text embeddings across many NLP and retrieval-adjacent tasks; BEIR is an information-retrieval benchmark suite.","target":"beir","type":"notEquivalentTo"},{"note":"MTEB is a benchmark suite for evaluating embedding models; an embedding is a vector representation.","target":"embedding","type":"notEquivalentTo"},{"note":"MTEB is a benchmark suite aggregating many tasks and datasets; a test collection is an evaluation resource pattern.","target":"test-collection","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"embedding","type":"related"},{"target":"reranking","type":"related"},{"target":"retrieval","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/mteb/","json":"https://id.searchplex.net/mteb.json","jsonld":"https://id.searchplex.net/mteb.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"MTEB includes retrieval and reranking tasks, but it is broader than IR evaluation alone because it also covers tasks such as classification, clustering, semantic textual similarity, summarization, pair classification, and bitext mining. It should not be treated as the same kind of benchmark suite as BEIR, which is focused on information retrieval.","status":"published","terms":[{"community":"ml-nlp","label":"MTEB Leaderboard","usage":"nearSynonym"},{"community":"industry","label":"Embedding Benchmark","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/mteb/"}
{"altLabels":["Multihop Retrieval"],"definition":"A retrieval task or pattern in which satisfying an information need requires multiple dependent retrieval steps or evidence gathered across several linked pieces of information.","derivedRelations":[{"target":"agentic-rag","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"deep-research","type":"related"},{"target":"hotpotqa","type":"related"},{"target":"web-browsing-agent","type":"related"}],"distinctions":[{"reason":"Multi-hop retrieval is a retrieval pattern; agentic retrieval is the broader agent-controlled loop that may perform multi-hop retrieval.","statement":"Multi-Hop Retrieval is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Multi-hop retrieval involves dependent evidence hops or subproblems; iterative retrieval is broader and includes any repeated retrieval attempts.","statement":"Multi-Hop Retrieval is not Iterative Retrieval.","target":"iterative-retrieval","targetLabel":"Iterative Retrieval","targetUri":"https://id.searchplex.net/iterative-retrieval/"},{"reason":"Multi-hop retrieval uses multiple dependent retrieval steps or evidence hops; multi-stage ranking is a retrieval-system architecture with successive ranking stages.","statement":"Multi-Hop Retrieval is not Multi-Stage Ranking.","target":"multi-stage-ranking","targetLabel":"Multi-Stage Ranking","targetUri":"https://id.searchplex.net/multi-stage-ranking/"},{"reason":"Retrieve-and-rerank retrieves a candidate set and then reranks it; multi-hop retrieval performs multiple dependent retrieval steps.","statement":"Multi-Hop Retrieval is not Retrieve-and-Rerank.","target":"retrieve-and-rerank","targetLabel":"Retrieve-and-Rerank","targetUri":"https://id.searchplex.net/retrieve-and-rerank/"},{"reason":"Multi-hop retrieval uses multiple dependent retrieval steps or evidence hops; single-shot retrieval uses one query or retrieval attempt without dependent follow-up steps.","statement":"Multi-Hop Retrieval is not Single-Shot Retrieval.","target":"single-shot-retrieval","targetLabel":"Single-Shot Retrieval","targetUri":"https://id.searchplex.net/single-shot-retrieval/"}],"id":"multi-hop-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"HotpotQA","type":"officialHomepage","url":"https://hotpotqa.github.io/"},{"label":"Agentic Retrieval-Augmented Generation","type":"canonicalSurvey","url":"https://arxiv.org/abs/2501.09136"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Multi-Hop Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"iterative-retrieval","type":"broader"},{"note":"Multi-hop retrieval is a retrieval pattern; agentic retrieval is the broader agent-controlled loop that may perform multi-hop retrieval.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Multi-hop retrieval involves dependent evidence hops or subproblems; iterative retrieval is broader and includes any repeated retrieval attempts.","target":"iterative-retrieval","type":"notEquivalentTo"},{"note":"Multi-hop retrieval uses multiple dependent retrieval steps or evidence hops; multi-stage ranking is a retrieval-system architecture with successive ranking stages.","target":"multi-stage-ranking","type":"notEquivalentTo"},{"note":"Retrieve-and-rerank retrieves a candidate set and then reranks it; multi-hop retrieval performs multiple dependent retrieval steps.","target":"retrieve-and-rerank","type":"notEquivalentTo"},{"note":"Multi-hop retrieval uses multiple dependent retrieval steps or evidence hops; single-shot retrieval uses one query or retrieval attempt without dependent follow-up steps.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"question-answering","type":"related"},{"target":"rag","type":"related"},{"target":"retrieval-planning","type":"related"},{"target":"single-shot-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/multi-hop-retrieval/","json":"https://id.searchplex.net/multi-hop-retrieval.json","jsonld":"https://id.searchplex.net/multi-hop-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Multi-hop retrieval is common in complex question answering and agentic RAG, where one retrieved result can inform the next retrieval step. It is distinct from multi-stage ranking, which describes a staged ranking architecture, and from retrieve-and-rerank, which is a particular pipeline pattern.","status":"published","terms":[{"community":"industry","label":"Multi-Step Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/multi-hop-retrieval/"}
{"altLabels":["Cascaded Ranking"],"definition":"Ranking architecture in which successive stages apply different retrieval or ranking methods to progressively refine results.","derivedRelations":[{"target":"first-stage-ranking","type":"hasPart"},{"target":"reranking","type":"hasPart"},{"target":"retrieve-and-rerank","type":"narrower"},{"note":"Multi-stage ranking is a retrieval-system architecture; agentic retrieval is a task-loop pattern in which an agent may call one or more retrieval systems repeatedly.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Multi-hop retrieval uses multiple dependent retrieval steps or evidence hops; multi-stage ranking is a retrieval-system architecture with successive ranking stages.","target":"multi-hop-retrieval","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"cross-encoder","type":"related"}],"id":"multi-stage-ranking","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Retrieve & Re-Rank","type":"officialHomepage","url":"https://www.sbert.net/examples/applications/retrieve_rerank/README.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Multi-Stage Ranking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"ranking","type":"broader"},{"target":"candidate-retrieval","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/multi-stage-ranking/","json":"https://id.searchplex.net/multi-stage-ranking.json","jsonld":"https://id.searchplex.net/multi-stage-ranking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Use for staged ranking pipelines rather than a single ranking model.","status":"published","terms":[{"community":"industry","label":"Ranking Cascade","usage":"communityPreferred"}],"uri":"https://id.searchplex.net/multi-stage-ranking/"}
{"definition":"Retrieval in which a query, an indexed item, or both are represented by multiple vectors rather than a single vector.","derivedRelations":[{"note":"Multimodal retrieval is about data modalities; multi-vector retrieval is about representing a query or item with multiple vectors.","target":"multimodal-retrieval","type":"notEquivalentTo"},{"target":"embedding","type":"related"},{"target":"multimodal-retrieval","type":"related"}],"distinctions":[{"reason":"Multi-vector retrieval is the broader representation pattern; ColBERT is a specific late-interaction model family.","statement":"Multi-Vector Retrieval is not ColBERT.","target":"colbert","targetLabel":"ColBERT","targetUri":"https://id.searchplex.net/colbert/"},{"reason":"Multi-vector retrieval is a representation pattern; late interaction is a scoring or interaction paradigm that often uses multi-vector representations.","statement":"Multi-Vector Retrieval is not Late Interaction.","target":"late-interaction","targetLabel":"Late Interaction","targetUri":"https://id.searchplex.net/late-interaction/"},{"reason":"Multi-vector retrieval uses multiple vectors for a query, item, or both; single-vector retrieval collapses each query or item to one vector.","statement":"Multi-Vector Retrieval is not Single-Vector Retrieval.","target":"single-vector-retrieval","targetLabel":"Single-Vector Retrieval","targetUri":"https://id.searchplex.net/single-vector-retrieval/"}],"id":"multi-vector-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Qdrant multivectors documentation","type":"officialHomepage","url":"https://qdrant.tech/documentation/concepts/vectors/#multivectors"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Multi-Vector Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Multi-vector retrieval is the broader representation pattern; ColBERT is a specific late-interaction model family.","target":"colbert","type":"notEquivalentTo"},{"note":"Multi-vector retrieval is a representation pattern; late interaction is a scoring or interaction paradigm that often uses multi-vector representations.","target":"late-interaction","type":"notEquivalentTo"},{"note":"Multi-vector retrieval uses multiple vectors for a query, item, or both; single-vector retrieval collapses each query or item to one vector.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"target":"colbert","type":"related"},{"target":"late-interaction","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/multi-vector-retrieval/","json":"https://id.searchplex.net/multi-vector-retrieval.json","jsonld":"https://id.searchplex.net/multi-vector-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Late interaction commonly uses multi-vector representations, but multi-vector retrieval is a broader representational pattern and does not by itself imply ColBERT-style interaction.","status":"published","uri":"https://id.searchplex.net/multi-vector-retrieval/"}
{"altLabels":["Multilingual Information Retrieval","MLIR"],"definition":"Retrieval over collections, queries, users, or results involving more than one language.","derivedRelations":[{"target":"cross-language-retrieval","type":"narrower"},{"target":"fire","type":"related"},{"target":"ntcir","type":"related"}],"distinctions":[{"reason":"Cross-language retrieval is the specific case where a query in one language retrieves relevant material in another; multilingual retrieval is the broader multi-language retrieval setting.","statement":"Multilingual Retrieval is not Cross-Language Retrieval.","target":"cross-language-retrieval","targetLabel":"Cross-Language Retrieval","targetUri":"https://id.searchplex.net/cross-language-retrieval/"}],"id":"multilingual-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Neural Approaches to Multilingual Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2209.01335"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Multilingual Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Cross-language retrieval is the specific case where a query in one language retrieves relevant material in another; multilingual retrieval is the broader multi-language retrieval setting.","target":"cross-language-retrieval","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/multilingual-retrieval/","json":"https://id.searchplex.net/multilingual-retrieval.json","jsonld":"https://id.searchplex.net/multilingual-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Multilingual retrieval is a broader setting than cross-language retrieval. It may include multiple monolingual indexes, mixed-language collections, multilingual result lists, language-specific processing, or cross-language retrieval where query and result languages differ.","status":"published","uri":"https://id.searchplex.net/multilingual-retrieval/"}
{"altLabels":["Multimodal Search","Multimedia Retrieval","Multimedia Information Retrieval"],"definition":"Retrieval involving information needs, indexed items, representations, or results that span more than one data modality, such as text, images, audio, video, tables, or structured metadata.","derivedRelations":[{"target":"cross-modal-retrieval","type":"narrower"},{"target":"image-retrieval","type":"narrower"},{"note":"TRECVID is an evaluation campaign for video analysis and retrieval; multimodal retrieval is a broader retrieval task family.","target":"trecvid","type":"notEquivalentTo"},{"target":"semantic-search","type":"related"},{"target":"tip-of-the-tongue-search","type":"related"},{"target":"trecvid","type":"related"}],"distinctions":[{"reason":"Multimodal retrieval is the broader family involving multiple modalities; cross-modal retrieval specifically crosses a modality boundary between query and retrieved items.","statement":"Multimodal Retrieval is not Cross-Modal Retrieval.","target":"cross-modal-retrieval","targetLabel":"Cross-Modal Retrieval","targetUri":"https://id.searchplex.net/cross-modal-retrieval/"},{"reason":"Image retrieval targets images; multimodal retrieval is broader and may involve images, text, video, audio, tables, or combinations of modalities.","statement":"Multimodal Retrieval is not Image Retrieval.","target":"image-retrieval","targetLabel":"Image Retrieval","targetUri":"https://id.searchplex.net/image-retrieval/"},{"reason":"Multimodal retrieval is about data modalities; multi-vector retrieval is about representing a query or item with multiple vectors.","statement":"Multimodal Retrieval is not Multi-Vector Retrieval.","target":"multi-vector-retrieval","targetLabel":"Multi-Vector Retrieval","targetUri":"https://id.searchplex.net/multi-vector-retrieval/"},{"reason":"Multimodal retrieval may be used inside multimodal RAG systems, but RAG is a retrieval-augmented generation pattern.","statement":"Multimodal Retrieval is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"multimodal-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Ask in Any Modality","type":"canonicalSurvey","url":"https://arxiv.org/abs/2502.08826"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Multimodal Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Multimodal retrieval is the broader family involving multiple modalities; cross-modal retrieval specifically crosses a modality boundary between query and retrieved items.","target":"cross-modal-retrieval","type":"notEquivalentTo"},{"note":"Image retrieval targets images; multimodal retrieval is broader and may involve images, text, video, audio, tables, or combinations of modalities.","target":"image-retrieval","type":"notEquivalentTo"},{"note":"Multimodal retrieval is about data modalities; multi-vector retrieval is about representing a query or item with multiple vectors.","target":"multi-vector-retrieval","type":"notEquivalentTo"},{"note":"Multimodal retrieval may be used inside multimodal RAG systems, but RAG is a retrieval-augmented generation pattern.","target":"rag","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"multi-vector-retrieval","type":"related"},{"target":"rag","type":"related"}],"representations":{"html":"https://id.searchplex.net/multimodal-retrieval/","json":"https://id.searchplex.net/multimodal-retrieval.json","jsonld":"https://id.searchplex.net/multimodal-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Multimodal retrieval is a broad family. It includes settings where multiple modalities are searched together, represented jointly, or used in query and result interpretation. It is distinct from cross-modal retrieval, where the query and target results are in different modalities, and from image retrieval, where the retrieved objects are images.","status":"published","terms":[{"community":"vendor","label":"Multimodal Vector Search","usage":"nearSynonym"},{"community":"industry","label":"Neural Multimodal Search","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/multimodal-retrieval/"}
{"altLabels":["NER"],"definition":"Identifying spans in text that denote named entities or entity-like mentions and assigning them entity types.","distinctions":[{"reason":"Named entity recognition identifies mention spans and types; entity linking maps those mentions to canonical entities.","statement":"Named Entity Recognition is not Entity Linking.","target":"entity-linking","targetLabel":"Entity Linking","targetUri":"https://id.searchplex.net/entity-linking/"},{"reason":"Query segmentation divides a query into meaningful spans; named entity recognition identifies spans that denote entities or entity-like mentions.","statement":"Named Entity Recognition is not Query Segmentation.","target":"query-segmentation","targetLabel":"Query Segmentation","targetUri":"https://id.searchplex.net/query-segmentation/"}],"id":"named-entity-recognition","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"A survey of named entity recognition and classification","type":"canonicalSurvey","url":"https://doi.org/10.1016/j.artint.2009.07.006"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Named Entity Recognition","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Named entity recognition identifies mention spans and types; entity linking maps those mentions to canonical entities.","target":"entity-linking","type":"notEquivalentTo"},{"note":"Query segmentation divides a query into meaningful spans; named entity recognition identifies spans that denote entities or entity-like mentions.","target":"query-segmentation","type":"notEquivalentTo"},{"note":"Often used as a query-understanding component when entities, brands, places, products, or attributes are expressed in a query.","target":"query-understanding","type":"partOf"},{"target":"entity-linking","type":"related"},{"target":"query-segmentation","type":"related"}],"representations":{"html":"https://id.searchplex.net/named-entity-recognition/","json":"https://id.searchplex.net/named-entity-recognition.json","jsonld":"https://id.searchplex.net/named-entity-recognition.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In search systems, named entity recognition may be applied to queries or documents as part of query understanding, entity extraction, routing, or catalog matching. It identifies and types mentions; it does not by itself link them to canonical entities.","slugAliases":["ner"],"status":"published","uri":"https://id.searchplex.net/named-entity-recognition/"}
{"altLabels":["NQ"],"artifactKind":"Dataset","definition":"A question-answering dataset based on natural user questions, used in retrieval and open-domain question-answering evaluation.","distinctions":[{"reason":"Natural Questions is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"Natural Questions is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"Natural Questions is a dataset; question answering is the task.","statement":"Natural Questions is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"}],"id":"natural-questions","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Natural Questions","type":"officialHomepage","url":"https://ai.google.com/research/NaturalQuestions"},{"label":"Natural Questions in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/nq"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Natural Questions","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Natural Questions is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"note":"Natural Questions is a dataset; question answering is the task.","target":"question-answering","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"question-answering","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/natural-questions/","json":"https://id.searchplex.net/natural-questions.json","jsonld":"https://id.searchplex.net/natural-questions.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Natural Questions appears in BEIR as an open-domain QA retrieval dataset. It should be distinguished from question answering as a task and from BEIR as a benchmark suite.","status":"published","uri":"https://id.searchplex.net/natural-questions/"}
{"altLabels":["nDCG","NDCG","Normalized Discounted Cumulative Gain"],"definition":"A ranking evaluation metric that discounts graded relevance by result position and normalizes by the ideal ranking.","derivedRelations":[{"note":"Qrels provide labels used to compute metrics such as nDCG; nDCG is the metric, not the judgment set.","target":"qrels","type":"notEquivalentTo"},{"note":"Relevance judgments provide labels used by metrics such as nDCG; nDCG is an evaluation metric, not a judgment.","target":"relevance-judgment","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"learning-to-rank","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"ranking","type":"related"}],"id":"ndcg","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Cumulated gain-based evaluation paper","type":"definingReference","url":"https://doi.org/10.1145/582415.582418"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"nDCG@K","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"cutoff-metric","type":"broader"},{"target":"graded-relevance-metric","type":"broader"},{"target":"rank-sensitive-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"reranking","type":"evaluates"},{"target":"average-precision","type":"related"},{"target":"mean-average-precision","type":"related"},{"target":"qrels","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"relevance-judgment","type":"related"}],"representations":{"html":"https://id.searchplex.net/ndcg/","json":"https://id.searchplex.net/ndcg.json","jsonld":"https://id.searchplex.net/ndcg.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"nDCG is commonly used when relevance is graded and rank position matters. It should not be confused with recall-oriented cutoff metrics.","status":"published","uri":"https://id.searchplex.net/ndcg/"}
{"altLabels":["NF Corpus"],"artifactKind":"Dataset","definition":"A biomedical and nutrition-focused information-retrieval test collection used for evaluating retrieval systems.","distinctions":[{"reason":"NFCorpus is a component evaluation dataset; BEIR is the benchmark suite that includes NFCorpus among other datasets.","statement":"NFCorpus is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"NFCorpus includes relevance judgments; qrels are the judgment files, not the dataset identity.","statement":"NFCorpus is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"}],"id":"nfcorpus","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"NFCorpus","type":"officialHomepage","url":"https://www.cl.uni-heidelberg.de/statnlpgroup/nfcorpus/"},{"label":"NFCorpus in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/nfcorpus"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"NFCorpus","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"NFCorpus is a component evaluation dataset; BEIR is the benchmark suite that includes NFCorpus among other datasets.","target":"beir","type":"notEquivalentTo"},{"note":"NFCorpus includes relevance judgments; qrels are the judgment files, not the dataset identity.","target":"qrels","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/nfcorpus/","json":"https://id.searchplex.net/nfcorpus.json","jsonld":"https://id.searchplex.net/nfcorpus.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"NFCorpus is one of the component evaluation datasets included in BEIR. It should be distinguished from BEIR as a benchmark suite and from qrels as the relevance-judgment component of the dataset.","status":"published","uri":"https://id.searchplex.net/nfcorpus/"}
{"altLabels":["NII Testbeds and Community for Information Access Research"],"definition":"The NII Testbeds and Community for Information Access Research, an evaluation program and conference series for information-access technologies.","distinctions":[{"reason":"NTCIR and CLEF are distinct evaluation programs with different communities, histories, and task emphases.","statement":"NTCIR is not CLEF.","target":"clef","targetLabel":"CLEF","targetUri":"https://id.searchplex.net/clef/"},{"reason":"NTCIR is an evaluation program and conference series; a test collection is an evaluation resource produced or used by specific tasks.","statement":"NTCIR is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"},{"reason":"NTCIR and TREC are distinct evaluation programs, though both organize shared information-retrieval evaluation tasks.","statement":"NTCIR is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"ntcir","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"NTCIR","type":"officialHomepage","url":"https://research.nii.ac.jp/ntcir/"},{"label":"Report on NTCIR-13","type":"canonicalSurvey","url":"https://sigir.org/files/forum/2018D/p034.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"NTCIR","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"NTCIR and CLEF are distinct evaluation programs with different communities, histories, and task emphases.","target":"clef","type":"notEquivalentTo"},{"note":"NTCIR is an evaluation program and conference series; a test collection is an evaluation resource produced or used by specific tasks.","target":"test-collection","type":"notEquivalentTo"},{"note":"NTCIR and TREC are distinct evaluation programs, though both organize shared information-retrieval evaluation tasks.","target":"trec","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"cross-language-retrieval","type":"related"},{"target":"fire","type":"related"},{"target":"multilingual-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"question-answering","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/ntcir/","json":"https://id.searchplex.net/ntcir.json","jsonld":"https://id.searchplex.net/ntcir.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"NTCIR has supported shared evaluation tasks for information retrieval, cross-language information access, question answering, summarization, patent retrieval, and related information-access problems, with strong historical focus on East Asian languages and contexts.","status":"published","uri":"https://id.searchplex.net/ntcir/"}
{"definition":"A retrieval task in which the retrievable units are passages or passage-like text segments rather than whole documents or other larger items.","derivedRelations":[{"note":"Document retrieval returns documents as the primary retrievable units; passage retrieval returns passages or passage-like segments.","target":"document-retrieval","type":"notEquivalentTo"},{"target":"corpus","type":"related"},{"target":"hotpotqa","type":"related"},{"target":"inex","type":"related"},{"target":"natural-questions","type":"related"},{"target":"question-answering","type":"related"}],"distinctions":[{"reason":"Chunking creates passage-like units; passage retrieval retrieves such units for a query or information need.","statement":"Passage Retrieval is not Chunking.","target":"chunking","targetLabel":"Chunking","targetUri":"https://id.searchplex.net/chunking/"},{"reason":"Dense Passage Retrieval is a specific dense neural method; passage retrieval is the broader task of retrieving passages.","statement":"Passage Retrieval is not Dense Passage Retrieval.","target":"dense-passage-retrieval","targetLabel":"Dense Passage Retrieval","targetUri":"https://id.searchplex.net/dense-passage-retrieval/"},{"reason":"Passage retrieval is often used inside RAG systems, but RAG is the broader retrieval-augmented generation pattern.","statement":"Passage Retrieval is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"passage-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Dense Passage Retrieval for Open-Domain Question Answering","type":"definingReference","url":"https://arxiv.org/abs/2004.04906"},{"label":"MS MARCO passage ranking benchmark","type":"officialHomepage","url":"https://ir-datasets.com/msmarco-passage.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Passage Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Chunking creates passage-like units; passage retrieval retrieves such units for a query or information need.","target":"chunking","type":"notEquivalentTo"},{"note":"Dense Passage Retrieval is a specific dense neural method; passage retrieval is the broader task of retrieving passages.","target":"dense-passage-retrieval","type":"notEquivalentTo"},{"note":"Passage retrieval is often used inside RAG systems, but RAG is the broader retrieval-augmented generation pattern.","target":"rag","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"chunking","type":"related"},{"target":"dense-passage-retrieval","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"rag","type":"related"}],"representations":{"html":"https://id.searchplex.net/passage-retrieval/","json":"https://id.searchplex.net/passage-retrieval.json","jsonld":"https://id.searchplex.net/passage-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Passage retrieval is common in open-domain question answering, RAG, and passage-ranking benchmarks. It is distinct from chunking, which creates or selects the units, and from passage ranking, which emphasizes ordering a set of passage candidates.","status":"published","terms":[{"community":"academic","label":"Passage Ranking","usage":"nearSynonym"},{"community":"industry","label":"Chunk Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/passage-retrieval/"}
{"artifactKind":"Software","definition":"Performance-optimized Late Interaction Driver, an efficient retrieval engine for accelerating late-interaction search, particularly with ColBERTv2-style representations.","distinctions":[{"reason":"PLAID is a serving and retrieval engine; ColBERT is a late-interaction model family.","statement":"PLAID is not ColBERT.","target":"colbert","targetLabel":"ColBERT","targetUri":"https://id.searchplex.net/colbert/"},{"reason":"PLAID is a retrieval engine and optimization strategy; ColBERTv2 is a model.","statement":"PLAID is not ColBERTv2.","target":"colbertv2","targetLabel":"ColBERTv2","targetUri":"https://id.searchplex.net/colbertv2/"}],"id":"plaid","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"PLAID paper","type":"definingReference","url":"https://doi.org/10.1145/3477495.3531835"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"PLAID","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"PLAID is a serving and retrieval engine; ColBERT is a late-interaction model family.","target":"colbert","type":"notEquivalentTo"},{"note":"PLAID is a retrieval engine and optimization strategy; ColBERTv2 is a model.","target":"colbertv2","type":"notEquivalentTo"},{"target":"colbertv2","type":"related"},{"target":"late-interaction","type":"related"},{"target":"maxsim","type":"related"}],"representations":{"html":"https://id.searchplex.net/plaid/","json":"https://id.searchplex.net/plaid.json","jsonld":"https://id.searchplex.net/plaid.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"PLAID is a serving and retrieval engine and optimization strategy, not a successor model to ColBERTv2.","status":"published","uri":"https://id.searchplex.net/plaid/"}
{"definition":"An evaluation-collection construction method in which documents retrieved by multiple systems or runs are combined into a pool for relevance assessment.","distinctions":[{"reason":"Pooling selects items to assess; qrels are the resulting structured relevance judgments.","statement":"Pooling is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"Pooling identifies candidates for assessment; a relevance judgment is an assessment of an item for a topic or query.","statement":"Pooling is not Relevance Judgment.","target":"relevance-judgment","targetLabel":"Relevance Judgment","targetUri":"https://id.searchplex.net/relevance-judgment/"},{"reason":"Pooling is one method used to build test collections; it is not the entire collection.","statement":"Pooling is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"pooling","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Overview of TREC-8","type":"canonicalSurvey","url":"https://trec.nist.gov/pubs/trec8/papers/overview_8.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Pooling","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Pooling selects items to assess; qrels are the resulting structured relevance judgments.","target":"qrels","type":"notEquivalentTo"},{"note":"Pooling identifies candidates for assessment; a relevance judgment is an assessment of an item for a topic or query.","target":"relevance-judgment","type":"notEquivalentTo"},{"note":"Pooling is one method used to build test collections; it is not the entire collection.","target":"test-collection","type":"notEquivalentTo"},{"target":"qrels","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/pooling/","json":"https://id.searchplex.net/pooling.json","jsonld":"https://id.searchplex.net/pooling.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Pooling is central to many TREC-style test collections. It is a way to choose items to judge, not the qrels file itself and not a relevance metric.","status":"published","uri":"https://id.searchplex.net/pooling/"}
{"definition":"A bias in retrieval, ranking, or recommendation where already popular items receive disproportionate exposure or preference.","distinctions":[{"reason":"Popularity bias concerns item popularity and exposure; position bias concerns the effect of display rank or position.","statement":"Popularity Bias is not Position Bias.","target":"position-bias","targetLabel":"Position Bias","targetUri":"https://id.searchplex.net/position-bias/"},{"reason":"Popularity bias favors popular items; relevance is the usefulness or topical fit of an item for a query or information need.","statement":"Popularity Bias is not Relevance.","target":"relevance","targetLabel":"Relevance","targetUri":"https://id.searchplex.net/relevance/"}],"id":"popularity-bias","kind":"Phenomenon","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"The Unfairness of Popularity Bias in Recommendation","type":"canonicalSurvey","url":"https://arxiv.org/abs/1907.13286"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Popularity Bias","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Popularity bias concerns item popularity and exposure; position bias concerns the effect of display rank or position.","target":"position-bias","type":"notEquivalentTo"},{"note":"Popularity bias favors popular items; relevance is the usefulness or topical fit of an item for a query or information need.","target":"relevance","type":"notEquivalentTo"},{"target":"position-bias","type":"related"},{"target":"ranking","type":"related"},{"target":"relevance","type":"related"}],"representations":{"html":"https://id.searchplex.net/popularity-bias/","json":"https://id.searchplex.net/popularity-bias.json","jsonld":"https://id.searchplex.net/popularity-bias.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Popularity bias can affect recommendations, search ranking, training data, and implicit feedback loops. It is not the same as relevance, although popularity can be used as one relevance or utility signal.","status":"published","uri":"https://id.searchplex.net/popularity-bias/"}
{"altLabels":["Rank Position Bias"],"definition":"A user-behavior bias in which items shown at higher positions receive more attention, clicks, or interaction because of their position, independent of their true relevance.","derivedRelations":[{"note":"Popularity bias concerns item popularity and exposure; position bias concerns the effect of display rank or position.","target":"popularity-bias","type":"notEquivalentTo"},{"target":"popularity-bias","type":"related"}],"distinctions":[{"reason":"Ranking orders items; position bias is a behavioral effect of the displayed order.","statement":"Position Bias is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"Position bias is caused by where an item is displayed; relevance is the item's usefulness or topical fit for the information need.","statement":"Position Bias is not Relevance.","target":"relevance","targetLabel":"Relevance","targetUri":"https://id.searchplex.net/relevance/"}],"id":"position-bias","kind":"Phenomenon","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Estimating Position Bias without Intrusive Interventions","type":"definingReference","url":"https://www.cs.cornell.edu/~adith/CfactSIGIR2019.pdf"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Position Bias","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Ranking orders items; position bias is a behavioral effect of the displayed order.","target":"ranking","type":"notEquivalentTo"},{"note":"Position bias is caused by where an item is displayed; relevance is the item's usefulness or topical fit for the information need.","target":"relevance","type":"notEquivalentTo"},{"target":"learning-to-rank","type":"related"},{"target":"ranking","type":"related"},{"target":"relevance","type":"related"}],"representations":{"html":"https://id.searchplex.net/position-bias/","json":"https://id.searchplex.net/position-bias.json","jsonld":"https://id.searchplex.net/position-bias.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Position bias affects click logs, implicit feedback, learning-to-rank data, and evaluation based on user interactions. It is closely related to presentation bias but should not be treated as relevance itself.","status":"published","terms":[{"community":"academic","label":"Presentation Bias","usage":"nearSynonym"},{"community":"academic","label":"Examination Bias","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/position-bias/"}
{"altLabels":["Precision-Oriented Metric","Precision-Based Metric"],"definition":"A family of retrieval or ranking evaluation metrics that emphasize the proportion of retrieved or ranked results that are relevant.","derivedRelations":[{"target":"average-precision","type":"narrower"},{"target":"mean-average-precision","type":"narrower"},{"target":"precision-at-k","type":"narrower"}],"distinctions":[{"reason":"Precision metrics emphasize the fraction of returned results that are relevant; recall metrics emphasize how much of the relevant set was retrieved.","statement":"Precision Metric is not Recall Metric.","target":"recall-metric","targetLabel":"Recall Metric","targetUri":"https://id.searchplex.net/recall-metric/"}],"id":"precision-metric","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Precision Metric","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Precision metrics emphasize the fraction of returned results that are relevant; recall metrics emphasize how much of the relevant set was retrieved.","target":"recall-metric","type":"notEquivalentTo"},{"target":"cutoff-metric","type":"related"},{"target":"ranking","type":"related"},{"target":"recall-metric","type":"related"}],"representations":{"html":"https://id.searchplex.net/precision-metric/","json":"https://id.searchplex.net/precision-metric.json","jsonld":"https://id.searchplex.net/precision-metric.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Precision-oriented metrics focus on result quality among returned items, often at a cutoff or across relevant ranks. They are distinct from recall-oriented metrics, which emphasize coverage of the relevant set.","status":"published","uri":"https://id.searchplex.net/precision-metric/"}
{"definition":"An evaluation metric measuring the fraction of the top K retrieved or ranked results that are relevant.","derivedRelations":[{"note":"Precision@K measures precision at a fixed cutoff; Average Precision aggregates precision at relevant ranks across the ranked list.","target":"average-precision","type":"notEquivalentTo"},{"note":"MAP averages rank-sensitive AP values across topics; Precision@K only measures precision at a fixed cutoff.","target":"mean-average-precision","type":"notEquivalentTo"},{"note":"Success@K is binary per query; Precision@K measures the fraction of top-K results that are relevant.","target":"success-at-k","type":"notEquivalentTo"}],"distinctions":[{"reason":"Precision@K measures how many top-K results are relevant; Recall@K measures how much of the relevant set was retrieved within K.","statement":"Precision@K is not Recall@K.","target":"recall-at-k","targetLabel":"Recall@K","targetUri":"https://id.searchplex.net/recall-at-k/"}],"id":"precision-at-k","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Precision@K","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"cutoff-metric","type":"broader"},{"target":"precision-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"Precision@K measures how many top-K results are relevant; Recall@K measures how much of the relevant set was retrieved within K.","target":"recall-at-k","type":"notEquivalentTo"},{"target":"average-precision","type":"related"},{"target":"mean-average-precision","type":"related"},{"target":"mrr","type":"related"},{"target":"ndcg","type":"related"},{"target":"qrels","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"success-at-k","type":"related"}],"representations":{"html":"https://id.searchplex.net/precision-at-k/","json":"https://id.searchplex.net/precision-at-k.json","jsonld":"https://id.searchplex.net/precision-at-k.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Precision@K emphasizes the quality of the first K results. It is distinct from Recall@K, which measures how much of the relevant set appears within the top K.","status":"published","uri":"https://id.searchplex.net/precision-at-k/"}
{"altLabels":["PRF"],"definition":"A relevance-feedback method that treats top-ranked results from an initial retrieval as if they were relevant and uses them to modify a subsequent retrieval query or model.","derivedRelations":[{"target":"iterative-retrieval","type":"related"},{"target":"query-expansion","type":"related"}],"distinctions":[{"reason":"Pseudo-relevance feedback is a method for using top-ranked results as feedback; query expansion is one task or outcome it may support.","statement":"Pseudo-Relevance Feedback is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"}],"id":"pseudo-relevance-feedback","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Relevance feedback and query expansion","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Pseudo-Relevance Feedback","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Pseudo-relevance feedback is a method for using top-ranked results as feedback; query expansion is one task or outcome it may support.","target":"query-expansion","type":"notEquivalentTo"},{"target":"query-drift","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-expansion","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/pseudo-relevance-feedback/","json":"https://id.searchplex.net/pseudo-relevance-feedback.json","jsonld":"https://id.searchplex.net/pseudo-relevance-feedback.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Pseudo-relevance feedback removes the requirement for explicit user relevance judgments. It is commonly used to derive query expansion terms or updated query models.","status":"published","uri":"https://id.searchplex.net/pseudo-relevance-feedback/"}
{"altLabels":["Qrel","Query Relevance Judgments"],"artifactKind":"Dataset","definition":"A structured set or file of query-document relevance judgments used to evaluate retrieval runs.","derivedRelations":[{"note":"BEIR includes datasets with relevance judgments; qrels are judgment files, not the benchmark suite itself.","target":"beir","type":"notEquivalentTo"},{"note":"CLEF labs may publish relevance judgments or qrels; qrels are judgment files, not the CLEF program itself.","target":"clef","type":"notEquivalentTo"},{"note":"A corpus contains retrievable items; qrels contain relevance judgments for query-item pairs.","target":"corpus","type":"notEquivalentTo"},{"note":"ir_datasets can expose qrels; qrels are relevance-judgment files, not the software package.","target":"ir-datasets","type":"notEquivalentTo"},{"note":"NFCorpus includes relevance judgments; qrels are the judgment files, not the dataset identity.","target":"nfcorpus","type":"notEquivalentTo"},{"note":"SciFact includes relevance or evidence judgments; qrels are judgment files, not the dataset identity.","target":"scifact","type":"notEquivalentTo"},{"note":"TREC may publish qrels for tracks; qrels are judgment files, not the TREC program itself.","target":"trec","type":"notEquivalentTo"},{"target":"arguana","type":"related"},{"target":"average-precision","type":"related"},{"target":"beir","type":"related"},{"target":"clef","type":"related"},{"target":"corpus","type":"related"},{"target":"fever","type":"related"},{"target":"fiqa-2018","type":"related"},{"target":"fire","type":"related"},{"target":"graded-relevance-metric","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"mean-average-precision","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"nfcorpus","type":"related"},{"target":"ntcir","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"rank-sensitive-metric","type":"related"},{"target":"scifact","type":"related"},{"target":"success-at-k","type":"related"},{"target":"trec-covid","type":"related"}],"distinctions":[{"reason":"LLM-as-judge is a method for producing or assisting labels; qrels are judgment data.","statement":"Qrels is not LLM-as-Judge.","target":"llm-as-judge","targetLabel":"LLM-as-Judge","targetUri":"https://id.searchplex.net/llm-as-judge/"},{"reason":"Qrels provide labels used to compute metrics such as nDCG; nDCG is the metric, not the judgment set.","statement":"Qrels is not nDCG@K.","target":"ndcg","targetLabel":"nDCG@K","targetUri":"https://id.searchplex.net/ndcg/"},{"reason":"Pooling is a method for selecting items to judge; qrels are the resulting structured judgment data.","statement":"Qrels is not Pooling.","target":"pooling","targetLabel":"Pooling","targetUri":"https://id.searchplex.net/pooling/"},{"reason":"Qrels are recorded labels about relevance; they are not the relevance relationship itself.","statement":"Qrels is not Relevance.","target":"relevance","targetLabel":"Relevance","targetUri":"https://id.searchplex.net/relevance/"},{"reason":"A qrels file contains many relevance judgments; an individual relevance judgment is one assessment.","statement":"Qrels is not Relevance Judgment.","target":"relevance-judgment","targetLabel":"Relevance Judgment","targetUri":"https://id.searchplex.net/relevance-judgment/"},{"reason":"Qrels are one component of a test collection, alongside resources such as a corpus and topics or queries.","statement":"Qrels is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"qrels","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC relevance judgments","type":"officialHomepage","url":"https://trec.nist.gov/data/qrels_eng/"},{"label":"trec_eval","type":"officialRepository","url":"https://github.com/usnistgov/trec_eval"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Qrels","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"LLM-as-judge is a method for producing or assisting labels; qrels are judgment data.","target":"llm-as-judge","type":"notEquivalentTo"},{"note":"Qrels provide labels used to compute metrics such as nDCG; nDCG is the metric, not the judgment set.","target":"ndcg","type":"notEquivalentTo"},{"note":"Pooling is a method for selecting items to judge; qrels are the resulting structured judgment data.","target":"pooling","type":"notEquivalentTo"},{"note":"Qrels are recorded labels about relevance; they are not the relevance relationship itself.","target":"relevance","type":"notEquivalentTo"},{"note":"A qrels file contains many relevance judgments; an individual relevance judgment is one assessment.","target":"relevance-judgment","type":"notEquivalentTo"},{"note":"Qrels are one component of a test collection, alongside resources such as a corpus and topics or queries.","target":"test-collection","type":"notEquivalentTo"},{"target":"test-collection","type":"partOf"},{"target":"llm-as-judge","type":"related"},{"target":"mrr","type":"related"},{"target":"ndcg","type":"related"},{"target":"pooling","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"relevance","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/qrels/","json":"https://id.searchplex.net/qrels.json","jsonld":"https://id.searchplex.net/qrels.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Qrels is TREC-style terminology for query relevance judgments. A qrels file usually records relevance labels for query-item pairs and is consumed by evaluation tools such as trec_eval. Qrels are not evaluation metrics and are not the same as the full test collection.","status":"published","uri":"https://id.searchplex.net/qrels/"}
{"altLabels":["Query Auto Completion","QAC"],"definition":"Predicting and ranking complete search queries from a partially typed query prefix.","derivedRelations":[{"note":"Spelling correction corrects a submitted or completed query; query autocomplete predicts completions from a partially typed prefix.","target":"spelling-correction","type":"notEquivalentTo"}],"distinctions":[{"reason":"Query autocomplete completes a typed prefix; query suggestion recommends related, next, or alternative queries and need not be prefix-completion.","statement":"Query Autocomplete is not Query Suggestion.","target":"query-suggestion","targetLabel":"Query Suggestion","targetUri":"https://id.searchplex.net/query-suggestion/"},{"reason":"Query autocomplete proposes completions for a query prefix; search-as-you-type updates actual search results while the user types.","statement":"Query Autocomplete is not Search-as-You-Type.","target":"search-as-you-type","targetLabel":"Search-as-You-Type","targetUri":"https://id.searchplex.net/search-as-you-type/"}],"id":"query-autocomplete","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"AmazonQAC: A Large-Scale, Naturalistic Query Autocomplete Dataset","type":"definingReference","url":"https://aclanthology.org/2024.emnlp-industry.78/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Autocomplete","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query autocomplete completes a typed prefix; query suggestion recommends related, next, or alternative queries and need not be prefix-completion.","target":"query-suggestion","type":"notEquivalentTo"},{"note":"Query autocomplete proposes completions for a query prefix; search-as-you-type updates actual search results while the user types.","target":"search-as-you-type","type":"notEquivalentTo"},{"note":"Often implemented near query understanding and query assistance components.","target":"query-understanding","type":"partOf"},{"target":"query-suggestion","type":"related"},{"target":"search-as-you-type","type":"related"},{"target":"session-search","type":"related"},{"target":"spelling-correction","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-autocomplete/","json":"https://id.searchplex.net/query-autocomplete.json","jsonld":"https://id.searchplex.net/query-autocomplete.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query autocomplete is prefix-conditioned: the user has begun typing a query and the system proposes likely completions. It is distinct from query suggestion, which may recommend related or subsequent queries without being limited to completing a current prefix.","slugAliases":["qac"],"status":"published","terms":[{"community":"industry","label":"Typeahead","usage":"nearSynonym"},{"community":"industry","label":"Search Autocomplete","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-autocomplete/"}
{"altLabels":["Web Query Classification","Query Categorization"],"definition":"Assigning a search query to one or more predefined classes, categories, intents, verticals, or other labels.","distinctions":[{"reason":"Query classification assigns labels or categories; query routing chooses a source, tool, retriever, model, or strategy.","statement":"Query Classification is not Query Routing.","target":"query-routing","targetLabel":"Query Routing","targetUri":"https://id.searchplex.net/query-routing/"},{"reason":"Query understanding is the broader interpretation workflow; query classification is one task within that workflow.","statement":"Query Classification is not Query Understanding.","target":"query-understanding","targetLabel":"Query Understanding","targetUri":"https://id.searchplex.net/query-understanding/"}],"id":"query-classification","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Building bridges for web query classification","type":"definingReference","url":"https://dl.acm.org/doi/10.1145/1390334.1390335"},{"label":"ORCAS-I: Queries Annotated with Intent using Weak Supervision","type":"definingReference","url":"https://arxiv.org/abs/2205.00926"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Classification","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query classification assigns labels or categories; query routing chooses a source, tool, retriever, model, or strategy.","target":"query-routing","type":"notEquivalentTo"},{"note":"Query understanding is the broader interpretation workflow; query classification is one task within that workflow.","target":"query-understanding","type":"notEquivalentTo"},{"target":"query-understanding","type":"partOf"},{"target":"query-routing","type":"related"},{"target":"ranking","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-classification/","json":"https://id.searchplex.net/query-classification.json","jsonld":"https://id.searchplex.net/query-classification.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query classification is broader than query intent classification. Depending on the system, the target classes may represent topic, product category, vertical, navigational intent, commercial intent, routing class, or another operational taxonomy.","status":"published","terms":[{"community":"academic","label":"Query Intent Classification","usage":"nearSynonym"},{"community":"commerce-search","label":"Category Prediction","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-classification/"}
{"altLabels":["Question Decomposition"],"definition":"Breaking a complex query or information need into smaller subqueries or subproblems that can be retrieved or answered separately.","derivedRelations":[{"target":"agentic-rag","type":"related"},{"target":"deep-research","type":"related"}],"distinctions":[{"reason":"Query decomposition breaks a complex information need into subqueries; query reformulation changes an existing query to better express or satisfy an information need.","statement":"Query Decomposition is not Query Reformulation.","target":"query-reformulation","targetLabel":"Query Reformulation","targetUri":"https://id.searchplex.net/query-reformulation/"},{"reason":"Query decomposition creates subqueries or subproblems; query routing chooses the handling path or retrieval resource.","statement":"Query Decomposition is not Query Routing.","target":"query-routing","targetLabel":"Query Routing","targetUri":"https://id.searchplex.net/query-routing/"},{"reason":"Query decomposition may be one step in a retrieval plan; retrieval planning chooses the overall retrieval strategy or sequence.","statement":"Query Decomposition is not Retrieval Planning.","target":"retrieval-planning","targetLabel":"Retrieval Planning","targetUri":"https://id.searchplex.net/retrieval-planning/"}],"id":"query-decomposition","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Decomposing Complex Queries for Tip-of-the-tongue Retrieval","type":"definingReference","url":"https://arxiv.org/abs/2305.15053"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Decomposition","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query decomposition breaks a complex information need into subqueries; query reformulation changes an existing query to better express or satisfy an information need.","target":"query-reformulation","type":"notEquivalentTo"},{"note":"Query decomposition creates subqueries or subproblems; query routing chooses the handling path or retrieval resource.","target":"query-routing","type":"notEquivalentTo"},{"note":"Query decomposition may be one step in a retrieval plan; retrieval planning chooses the overall retrieval strategy or sequence.","target":"retrieval-planning","type":"notEquivalentTo"},{"target":"query-understanding","type":"partOf"},{"target":"agentic-retrieval","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"question-answering","type":"related"},{"target":"retrieval-planning","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-decomposition/","json":"https://id.searchplex.net/query-decomposition.json","jsonld":"https://id.searchplex.net/query-decomposition.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query decomposition is common in multi-hop question answering, agentic retrieval, and complex search tasks. It is distinct from query reformulation, which changes a query, and from query routing, which decides where or how a query should be handled.","status":"published","terms":[{"community":"industry","label":"Subquery Generation","usage":"nearSynonym"},{"community":"ml-nlp","label":"Sub-Question Generation","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-decomposition/"}
{"definition":"A failure mode in which query modification or feedback moves the retrieval representation away from the user's intended information need.","distinctions":[{"reason":"Query drift is a harmful shift in query meaning; vocabulary mismatch is a gap between query and document wording.","statement":"Query Drift is not Vocabulary Mismatch.","target":"vocabulary-mismatch","targetLabel":"Vocabulary Mismatch","targetUri":"https://id.searchplex.net/vocabulary-mismatch/"}],"id":"query-drift","kind":"Phenomenon","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Query expansion techniques for information retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/1708.00247"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Drift","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query drift is a harmful shift in query meaning; vocabulary mismatch is a gap between query and document wording.","target":"vocabulary-mismatch","type":"notEquivalentTo"},{"target":"pseudo-relevance-feedback","type":"related"},{"target":"query-expansion","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-drift/","json":"https://id.searchplex.net/query-drift.json","jsonld":"https://id.searchplex.net/query-drift.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query drift is often discussed around query expansion and pseudo-relevance feedback, where added evidence can change the effective query meaning.","status":"published","uri":"https://id.searchplex.net/query-drift/"}
{"altLabels":["Automatic Query Expansion","AQE"],"definition":"Enriching a query with additional retrieval evidence intended to improve retrieval effectiveness.","derivedRelations":[{"note":"Pseudo-relevance feedback is a method for using top-ranked results as feedback; query expansion is one task or outcome it may support.","target":"pseudo-relevance-feedback","type":"notEquivalentTo"},{"note":"Query expansion adds retrieval evidence; query rewriting may transform, normalize, constrain, or restructure a query in other ways.","target":"query-rewriting","type":"notEquivalentTo"},{"note":"Query expansion is one technique for addressing vocabulary mismatch; vocabulary mismatch is the underlying terminology gap.","target":"vocabulary-mismatch","type":"notEquivalentTo"},{"target":"query-drift","type":"related"},{"target":"pseudo-relevance-feedback","type":"uses"}],"distinctions":[{"reason":"Query expansion enriches the query representation; document expansion enriches the indexed document representation.","statement":"Query Expansion is not Document Expansion.","target":"document-expansion","targetLabel":"Document Expansion","targetUri":"https://id.searchplex.net/document-expansion/"},{"reason":"Learned sparse retrieval may produce expansion-like representations, but it is a retrieval representation family rather than the query-expansion task.","statement":"Query Expansion is not Learned Sparse Retrieval.","target":"learned-sparse-retrieval","targetLabel":"Learned Sparse Retrieval","targetUri":"https://id.searchplex.net/learned-sparse-retrieval/"},{"reason":"Query expansion adds retrieval evidence; query relaxation loosens or removes matching requirements.","statement":"Query Expansion is not Query Relaxation.","target":"query-relaxation","targetLabel":"Query Relaxation","targetUri":"https://id.searchplex.net/query-relaxation/"},{"reason":"Query understanding is the broader interpretation workflow; query expansion is one possible transformation within that workflow.","statement":"Query Expansion is not Query Understanding.","target":"query-understanding","targetLabel":"Query Understanding","targetUri":"https://id.searchplex.net/query-understanding/"},{"reason":"SPLADE is a learned sparse model family with expansion-like behavior, not the task of query expansion.","statement":"Query Expansion is not SPLADE.","target":"splade","targetLabel":"SPLADE","targetUri":"https://id.searchplex.net/splade/"}],"id":"query-expansion","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Relevance feedback and query expansion","type":"definingReference","url":"https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html"}],"mappings":[{"relation":"closeMatch","scheme":"wikidata","uri":"https://www.wikidata.org/entity/Q3500136"},{"relation":"relatedMatch","scheme":"acm-ccs-2012","uri":"https://dl.acm.org/ccs/ccs_flat.cfm#10002951.10003317"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Expansion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"query-transformation","type":"broader"},{"note":"Query expansion enriches the query representation; document expansion enriches the indexed document representation.","target":"document-expansion","type":"notEquivalentTo"},{"note":"Learned sparse retrieval may produce expansion-like representations, but it is a retrieval representation family rather than the query-expansion task.","target":"learned-sparse-retrieval","type":"notEquivalentTo"},{"note":"Query expansion adds retrieval evidence; query relaxation loosens or removes matching requirements.","target":"query-relaxation","type":"notEquivalentTo"},{"note":"Query understanding is the broader interpretation workflow; query expansion is one possible transformation within that workflow.","target":"query-understanding","type":"notEquivalentTo"},{"note":"SPLADE is a learned sparse model family with expansion-like behavior, not the task of query expansion.","target":"splade","type":"notEquivalentTo"},{"note":"Often implemented as a query-understanding workflow component.","target":"query-understanding","type":"partOf"},{"target":"document-expansion","type":"related"},{"target":"learned-sparse-retrieval","type":"related"},{"target":"pseudo-relevance-feedback","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-relaxation","type":"related"},{"target":"query-rewriting","type":"related"},{"target":"splade","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-expansion/","json":"https://id.searchplex.net/query-expansion.json","jsonld":"https://id.searchplex.net/query-expansion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Covers expansion of the query representation rather than ranking-stage relevance estimation.","slugAliases":["aqe"],"status":"published","uri":"https://id.searchplex.net/query-expansion/"}
{"definition":"The set of operations that analyze, interpret, transform, or otherwise prepare a query for execution by a retrieval system.","derivedRelations":[{"target":"query-understanding","type":"hasPart"}],"distinctions":[{"reason":"Query processing is the broader preparation pipeline; query understanding is the interpretation component within that pipeline.","statement":"Query Processing is not Query Understanding.","target":"query-understanding","targetLabel":"Query Understanding","targetUri":"https://id.searchplex.net/query-understanding/"}],"id":"query-processing","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Processing Boolean queries","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/processing-boolean-queries-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Processing","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query processing is the broader preparation pipeline; query understanding is the interpretation component within that pipeline.","target":"query-understanding","type":"notEquivalentTo"},{"target":"query-transformation","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-processing/","json":"https://id.searchplex.net/query-processing.json","jsonld":"https://id.searchplex.net/query-processing.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query processing is an umbrella term. Depending on the system, it may include normalization, query understanding, classification, expansion, rewriting, field assignment, routing, and retrieval planning.","status":"published","uri":"https://id.searchplex.net/query-processing/"}
{"definition":"Changing an existing query in order to better express or satisfy an information need.","derivedRelations":[{"note":"Query reformulation changes a query; iterative retrieval is the broader repeated retrieval pattern that may use reformulation between attempts.","target":"iterative-retrieval","type":"notEquivalentTo"},{"note":"Query decomposition breaks a complex information need into subqueries; query reformulation changes an existing query to better express or satisfy an information need.","target":"query-decomposition","type":"notEquivalentTo"},{"note":"Query reformulation is the act or process of changing a query; query suggestion is the system task of proposing candidate queries to the user.","target":"query-suggestion","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"conversational-search","type":"related"},{"target":"interactive-search","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"pseudo-relevance-feedback","type":"related"},{"target":"query-suggestion","type":"related"},{"target":"session-search","type":"related"},{"target":"tip-of-the-tongue-search","type":"related"}],"id":"query-reformulation","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Query expansion techniques for information retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/1708.00247"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Reformulation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"query-expansion","type":"related"},{"target":"query-rewriting","type":"related"},{"target":"query-transformation","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-reformulation/","json":"https://id.searchplex.net/query-reformulation.json","jsonld":"https://id.searchplex.net/query-reformulation.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In interactive IR, query reformulation often describes a user modifying a query during a search session. The term is also used for automatic system-driven transformations, where it overlaps with query rewriting.","status":"published","uri":"https://id.searchplex.net/query-reformulation/"}
{"altLabels":["Automatic Query Relaxation"],"definition":"Transforming a query or its matching requirements to make retrieval less restrictive, often to avoid zero or low-result searches.","derivedRelations":[{"note":"Spelling correction changes likely misspelled terms; query relaxation loosens constraints or matching requirements.","target":"spelling-correction","type":"notEquivalentTo"},{"target":"spelling-correction","type":"related"}],"distinctions":[{"reason":"Exact match requires literal matching; query relaxation may loosen exact-match requirements when they are too restrictive.","statement":"Query Relaxation is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"Query expansion adds retrieval evidence or related terms; query relaxation loosens or removes matching requirements.","statement":"Query Relaxation is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"},{"reason":"Query rewriting is the broader transformation family; query relaxation is the specific transformation that makes matching less restrictive.","statement":"Query Relaxation is not Query Rewriting.","target":"query-rewriting","targetLabel":"Query Rewriting","targetUri":"https://id.searchplex.net/query-rewriting/"},{"reason":"Result filtering restricts eligible results; query relaxation broadens eligibility or weakens matching criteria.","statement":"Query Relaxation is not Result Filtering.","target":"result-filtering","targetLabel":"Result Filtering","targetUri":"https://id.searchplex.net/result-filtering/"}],"id":"query-relaxation","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Adapting information retrieval systems to user queries","type":"definingReference","url":"https://www.sciencedirect.com/science/article/pii/S0306457308000514"},{"label":"Bloomreach query relaxation documentation","type":"officialHomepage","url":"https://documentation.bloomreach.com/discovery/docs/query-relaxation"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Relaxation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"query-transformation","type":"broader"},{"note":"Exact match requires literal matching; query relaxation may loosen exact-match requirements when they are too restrictive.","target":"exact-match","type":"notEquivalentTo"},{"note":"Query expansion adds retrieval evidence or related terms; query relaxation loosens or removes matching requirements.","target":"query-expansion","type":"notEquivalentTo"},{"note":"Query rewriting is the broader transformation family; query relaxation is the specific transformation that makes matching less restrictive.","target":"query-rewriting","type":"notEquivalentTo"},{"note":"Result filtering restricts eligible results; query relaxation broadens eligibility or weakens matching criteria.","target":"result-filtering","type":"notEquivalentTo"},{"target":"query-understanding","type":"partOf"},{"target":"exact-match","type":"related"},{"target":"query-expansion","type":"related"},{"target":"query-rewriting","type":"related"},{"target":"result-filtering","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-relaxation/","json":"https://id.searchplex.net/query-relaxation.json","jsonld":"https://id.searchplex.net/query-relaxation.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query relaxation may drop, down-weight, or loosen terms, constraints, exact matches, or matching criteria. It is often discussed as the counterpart of query expansion: expansion adds evidence, while relaxation weakens requirements.","status":"published","terms":[{"community":"academic","label":"AQR","usage":"abbreviation"},{"community":"industry","label":"Query Broadening","usage":"nearSynonym"},{"community":"industry","label":"Term Dropping","usage":"nearSynonym"},{"community":"commerce-search","label":"Partial Matching","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-relaxation/"}
{"definition":"Automatically transforming an input query into another query or query representation intended to improve retrieval or execution.","derivedRelations":[{"note":"Spelling correction is a specific query-assistance task; query rewriting is the broader family of query transformations.","target":"spelling-correction","type":"notEquivalentTo"}],"distinctions":[{"reason":"Query expansion adds retrieval evidence; query rewriting may transform, normalize, constrain, or restructure a query in other ways.","statement":"Query Rewriting is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"},{"reason":"Query rewriting is the broader transformation family; query relaxation specifically loosens or removes matching requirements.","statement":"Query Rewriting is not Query Relaxation.","target":"query-relaxation","targetLabel":"Query Relaxation","targetUri":"https://id.searchplex.net/query-relaxation/"},{"reason":"Query understanding interprets the query; query rewriting changes the query or query representation.","statement":"Query Rewriting is not Query Understanding.","target":"query-understanding","targetLabel":"Query Understanding","targetUri":"https://id.searchplex.net/query-understanding/"}],"id":"query-rewriting","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Crafting the Path: Robust Query Rewriting for Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2407.12529"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Rewriting","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"query-transformation","type":"broader"},{"note":"Query expansion adds retrieval evidence; query rewriting may transform, normalize, constrain, or restructure a query in other ways.","target":"query-expansion","type":"notEquivalentTo"},{"note":"Query rewriting is the broader transformation family; query relaxation specifically loosens or removes matching requirements.","target":"query-relaxation","type":"notEquivalentTo"},{"note":"Query understanding interprets the query; query rewriting changes the query or query representation.","target":"query-understanding","type":"notEquivalentTo"},{"target":"query-expansion","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-relaxation","type":"related"},{"target":"query-understanding","type":"related"},{"target":"spelling-correction","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-rewriting/","json":"https://id.searchplex.net/query-rewriting.json","jsonld":"https://id.searchplex.net/query-rewriting.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Depending on the system, rewriting may include spelling correction, normalization, expansion, reduction, phrase construction, structural changes, filters, or field-aware transformations. Usage is broader in some search-engine documentation than in academic IR.","status":"published","uri":"https://id.searchplex.net/query-rewriting/"}
{"definition":"Choosing where or how a query should be handled, such as selecting a source, index, retriever, tool, model, strategy, or no-retrieval path.","derivedRelations":[{"note":"Query decomposition creates subqueries or subproblems; query routing chooses the handling path or retrieval resource.","target":"query-decomposition","type":"notEquivalentTo"},{"note":"Retrieval planning chooses an overall retrieval strategy or sequence; query routing chooses a handling path, source, retriever, tool, or model for a query.","target":"retrieval-planning","type":"notEquivalentTo"},{"note":"Routing in classic IR uses standing queries or profiles over incoming documents; query routing chooses where or how a live query should be handled.","target":"routing","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"retrieval-planning","type":"related"}],"distinctions":[{"reason":"Query routing chooses a path for a query; agentic retrieval describes an agent-controlled retrieval loop that may include routing decisions.","statement":"Query Routing is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Query routing may be used inside federated search for source selection, but federated search is the broader multi-source retrieval pattern.","statement":"Query Routing is not Federated Search.","target":"federated-search","targetLabel":"Federated Search","targetUri":"https://id.searchplex.net/federated-search/"},{"reason":"Query classification assigns a query to labels or categories; query routing uses query signals to choose a handling path, source, retriever, model, or strategy.","statement":"Query Routing is not Query Classification.","target":"query-classification","targetLabel":"Query Classification","targetUri":"https://id.searchplex.net/query-classification/"}],"id":"query-routing","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Distributed Information Retrieval","type":"canonicalSurvey","url":"https://www.cs.cmu.edu/~callan/Papers/adcs.pdf"},{"label":"Efficient Federated Search for Retrieval-Augmented Generation","type":"definingReference","url":"https://arxiv.org/abs/2502.19280"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Routing","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query routing chooses a path for a query; agentic retrieval describes an agent-controlled retrieval loop that may include routing decisions.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Query routing may be used inside federated search for source selection, but federated search is the broader multi-source retrieval pattern.","target":"federated-search","type":"notEquivalentTo"},{"note":"Query classification assigns a query to labels or categories; query routing uses query signals to choose a handling path, source, retriever, model, or strategy.","target":"query-classification","type":"notEquivalentTo"},{"note":"Often uses query-understanding signals to decide which source, retriever, model, or strategy should handle the query.","target":"query-understanding","type":"partOf"},{"target":"agentic-retrieval","type":"related"},{"target":"federated-search","type":"related"},{"target":"query-classification","type":"related"},{"target":"rag","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-routing/","json":"https://id.searchplex.net/query-routing.json","jsonld":"https://id.searchplex.net/query-routing.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query routing uses information about the query to make a control-flow decision. In distributed IR and federated search, it often means selecting sources or servers. In RAG and agentic systems, it may mean selecting among retrievers, tools, indexes, models, or retrieval strategies.","status":"published","terms":[{"community":"academic","label":"Source Selection","usage":"nearSynonym"},{"community":"academic","label":"Resource Selection","usage":"nearSynonym"},{"community":"industry","label":"Router","usage":"informal"}],"uri":"https://id.searchplex.net/query-routing/"}
{"definition":"Dividing a search query into meaningful single-token or multi-token spans, optionally with broad semantic or operational types.","distinctions":[{"reason":"Query segmentation splits a short query into meaningful spans; chunking usually splits larger content into retrievable or processable units.","statement":"Query Segmentation is not Chunking.","target":"chunking","targetLabel":"Chunking","targetUri":"https://id.searchplex.net/chunking/"},{"reason":"Query segmentation divides the query into meaningful spans; named entity recognition identifies spans that denote entities or entity-like mentions.","statement":"Query Segmentation is not Named Entity Recognition.","target":"named-entity-recognition","targetLabel":"Named Entity Recognition","targetUri":"https://id.searchplex.net/named-entity-recognition/"}],"id":"query-segmentation","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"QueryNER: Segmentation of E-commerce Queries","type":"definingReference","url":"https://aclanthology.org/2024.lrec-main.1178/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Segmentation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query segmentation splits a short query into meaningful spans; chunking usually splits larger content into retrievable or processable units.","target":"chunking","type":"notEquivalentTo"},{"note":"Query segmentation divides the query into meaningful spans; named entity recognition identifies spans that denote entities or entity-like mentions.","target":"named-entity-recognition","type":"notEquivalentTo"},{"target":"query-understanding","type":"partOf"},{"target":"chunking","type":"related"},{"target":"named-entity-recognition","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-segmentation/","json":"https://id.searchplex.net/query-segmentation.json","jsonld":"https://id.searchplex.net/query-segmentation.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query segmentation identifies meaningful spans in a query. It is not the same as named entity recognition, because not every useful query segment is a named entity or entity-like mention.","status":"published","uri":"https://id.searchplex.net/query-segmentation/"}
{"altLabels":["Search Query Suggestion"],"definition":"Generating and ranking related, alternative, or subsequent search queries intended to help a user continue, refine, or redirect a search.","derivedRelations":[{"note":"Spelling correction proposes corrected forms of the user's query; query suggestion recommends related, alternative, or subsequent queries.","target":"spelling-correction","type":"notEquivalentTo"}],"distinctions":[{"reason":"Query suggestion recommends related, next, or alternative queries; query autocomplete completes the query prefix being typed.","statement":"Query Suggestion is not Query Autocomplete.","target":"query-autocomplete","targetLabel":"Query Autocomplete","targetUri":"https://id.searchplex.net/query-autocomplete/"},{"reason":"Query reformulation is the act or process of changing a query; query suggestion is the system task of proposing candidate queries to the user.","statement":"Query Suggestion is not Query Reformulation.","target":"query-reformulation","targetLabel":"Query Reformulation","targetUri":"https://id.searchplex.net/query-reformulation/"},{"reason":"Query suggestion proposes related or alternative queries; search-as-you-type updates retrieved results during query entry.","statement":"Query Suggestion is not Search-as-You-Type.","target":"search-as-you-type","targetLabel":"Search-as-You-Type","targetUri":"https://id.searchplex.net/search-as-you-type/"}],"id":"query-suggestion","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Context-aware query suggestion by mining click-through and session data","type":"definingReference","url":"https://doi.org/10.1145/1571941.1571945"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Suggestion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query suggestion recommends related, next, or alternative queries; query autocomplete completes the query prefix being typed.","target":"query-autocomplete","type":"notEquivalentTo"},{"note":"Query reformulation is the act or process of changing a query; query suggestion is the system task of proposing candidate queries to the user.","target":"query-reformulation","type":"notEquivalentTo"},{"note":"Query suggestion proposes related or alternative queries; search-as-you-type updates retrieved results during query entry.","target":"search-as-you-type","type":"notEquivalentTo"},{"note":"Often implemented as a query-assistance component near query understanding.","target":"query-understanding","type":"partOf"},{"target":"query-autocomplete","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"search-as-you-type","type":"related"},{"target":"session-search","type":"related"},{"target":"spelling-correction","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-suggestion/","json":"https://id.searchplex.net/query-suggestion.json","jsonld":"https://id.searchplex.net/query-suggestion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Query suggestion may use query logs, session context, click behavior, current results, or user context. It is distinct from query autocomplete, which completes a partially typed prefix.","status":"published","terms":[{"community":"industry","label":"Recommended Searches","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-suggestion/"}
{"definition":"Changing a query representation before retrieval or ranking while preserving or clarifying the user's information need.","derivedRelations":[{"target":"query-expansion","type":"narrower"},{"target":"query-relaxation","type":"narrower"},{"target":"query-rewriting","type":"narrower"}],"id":"query-transformation","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Crafting the Path: Robust Query Rewriting for Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2407.12529"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Transformation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"query-understanding","type":"partOf"},{"target":"query-processing","type":"related"},{"target":"query-reformulation","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-transformation/","json":"https://id.searchplex.net/query-transformation.json","jsonld":"https://id.searchplex.net/query-transformation.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Broader than query expansion; includes rewriting, normalization, decomposition, and related transformations.","status":"published","terms":[{"community":"industry","label":"Query Rewrite","usage":"informal"}],"uri":"https://id.searchplex.net/query-transformation/"}
{"altLabels":["Query Interpretation"],"definition":"Interpreting a user's query to infer intent, structure, entities, constraints, and useful retrieval signals.","derivedRelations":[{"note":"Often used in query understanding when recognized query mentions must be mapped to catalog, graph, or knowledge-base entities.","target":"entity-linking","type":"hasPart"},{"note":"Often used as a query-understanding component when entities, brands, places, products, or attributes are expressed in a query.","target":"named-entity-recognition","type":"hasPart"},{"note":"Often implemented near query understanding and query assistance components.","target":"query-autocomplete","type":"hasPart"},{"target":"query-classification","type":"hasPart"},{"target":"query-decomposition","type":"hasPart"},{"note":"Often implemented as a query-understanding workflow component.","target":"query-expansion","type":"hasPart"},{"target":"query-relaxation","type":"hasPart"},{"note":"Often uses query-understanding signals to decide which source, retriever, model, or strategy should handle the query.","target":"query-routing","type":"hasPart"},{"target":"query-segmentation","type":"hasPart"},{"note":"Often implemented as a query-assistance component near query understanding.","target":"query-suggestion","type":"hasPart"},{"target":"query-transformation","type":"hasPart"},{"target":"spelling-correction","type":"hasPart"},{"note":"Query understanding is the broader interpretation workflow; query classification is one task within that workflow.","target":"query-classification","type":"notEquivalentTo"},{"note":"Query processing is the broader preparation pipeline; query understanding is the interpretation component within that pipeline.","target":"query-processing","type":"notEquivalentTo"},{"note":"Search-as-you-type is an interaction pattern; query understanding is the interpretation of the query.","target":"search-as-you-type","type":"notEquivalentTo"},{"target":"diversified-retrieval","type":"related"},{"target":"search-as-you-type","type":"related"},{"target":"tip-of-the-tongue-search","type":"related"}],"distinctions":[{"reason":"Query understanding interprets intent and structure; query expansion enriches the query with additional retrieval evidence.","statement":"Query Understanding is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"},{"reason":"Query rewriting is a transformation step that may use query-understanding output.","statement":"Query Understanding is not Query Rewriting.","target":"query-rewriting","targetLabel":"Query Rewriting","targetUri":"https://id.searchplex.net/query-rewriting/"}],"id":"query-understanding","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Query understanding survey context","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000019"}],"mappings":[{"relation":"closeMatch","scheme":"wikidata","uri":"https://www.wikidata.org/entity/Q726877"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Query Understanding","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Query understanding interprets intent and structure; query expansion enriches the query with additional retrieval evidence.","target":"query-expansion","type":"notEquivalentTo"},{"note":"Query rewriting is a transformation step that may use query-understanding output.","target":"query-rewriting","type":"notEquivalentTo"},{"target":"query-processing","type":"partOf"},{"target":"query-rewriting","type":"related"}],"representations":{"html":"https://id.searchplex.net/query-understanding/","json":"https://id.searchplex.net/query-understanding.json","jsonld":"https://id.searchplex.net/query-understanding.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Covers analysis before retrieval rather than the retrieval or ranking stages themselves.","status":"published","terms":[{"community":"industry","label":"Query Analysis","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/query-understanding/"}
{"altLabels":["QA","Question-Answering"],"definition":"A task in which a system returns an answer, answer-bearing snippet, or response to a natural-language question rather than only a ranked list of documents.","derivedRelations":[{"note":"Question answering returns an answer to a question; deep research often produces a synthesized, cited research result after multiple search and reasoning steps.","target":"deep-research","type":"notEquivalentTo"},{"note":"FiQA-2018 is a dataset used for financial QA/retrieval evaluation; question answering is the task.","target":"fiqa-2018","type":"notEquivalentTo"},{"note":"HotpotQA is a dataset; question answering is the task.","target":"hotpotqa","type":"notEquivalentTo"},{"note":"Natural Questions is a dataset; question answering is the task.","target":"natural-questions","type":"notEquivalentTo"},{"note":"Tip-of-the-tongue search identifies an item from partial clues; question answering returns an answer or answer-bearing response.","target":"tip-of-the-tongue-search","type":"notEquivalentTo"},{"target":"deep-research","type":"related"},{"target":"fiqa-2018","type":"related"},{"target":"hotpotqa","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"natural-questions","type":"related"},{"target":"ntcir","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"tip-of-the-tongue-search","type":"related"}],"distinctions":[{"reason":"Question answering aims to return an answer or answer-bearing response; document retrieval returns documents for a query or information need.","statement":"Question Answering is not Document Retrieval.","target":"document-retrieval","targetLabel":"Document Retrieval","targetUri":"https://id.searchplex.net/document-retrieval/"},{"reason":"Question answering is a task objective; grounding concerns whether an answer or generated response is supported by source material.","statement":"Question Answering is not Grounding.","target":"grounding","targetLabel":"Grounding","targetUri":"https://id.searchplex.net/grounding/"},{"reason":"Question answering is a task; RAG is a retrieval-augmented generation pattern that can be used for question answering and other tasks.","statement":"Question Answering is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"question-answering","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC Question Answering Data","type":"officialHomepage","url":"https://trec.nist.gov/data/qa.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Question Answering","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Question answering aims to return an answer or answer-bearing response; document retrieval returns documents for a query or information need.","target":"document-retrieval","type":"notEquivalentTo"},{"note":"Question answering is a task objective; grounding concerns whether an answer or generated response is supported by source material.","target":"grounding","type":"notEquivalentTo"},{"note":"Question answering is a task; RAG is a retrieval-augmented generation pattern that can be used for question answering and other tasks.","target":"rag","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"grounding","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"rag","type":"related"}],"representations":{"html":"https://id.searchplex.net/question-answering/","json":"https://id.searchplex.net/question-answering.json","jsonld":"https://id.searchplex.net/question-answering.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Question answering may use retrieval internally, especially in open-domain QA and RAG systems, but the task objective is answer production. TREC question-answering evaluations historically emphasized returning short answers or snippets for fact-based questions.","status":"published","terms":[{"community":"industry","label":"Answer Search","usage":"nearSynonym"},{"community":"ml-nlp","label":"Answer Generation","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/question-answering/"}
{"definition":"A family of fusion methods that combine multiple ranked lists using rank positions or rank-derived evidence.","derivedRelations":[{"target":"reciprocal-rank-fusion","type":"narrower"},{"note":"RRF is one specific rank-fusion method, not the whole rank-fusion family.","target":"reciprocal-rank-fusion","type":"notEquivalentTo"},{"target":"hybrid-retrieval","type":"related"},{"target":"score-normalization","type":"related"}],"distinctions":[{"reason":"Rank fusion combines rank positions or rank-derived evidence; score fusion combines numerical scores.","statement":"Rank Fusion is not Score Fusion.","target":"score-fusion","targetLabel":"Score Fusion","targetUri":"https://id.searchplex.net/score-fusion/"}],"id":"rank-fusion","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Reciprocal Rank Fusion outperforms Condorcet and individual Rank Learning Methods","type":"definingReference","url":"https://doi.org/10.1145/1571941.1572114"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Rank Fusion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Rank fusion combines rank positions or rank-derived evidence; score fusion combines numerical scores.","target":"score-fusion","type":"notEquivalentTo"},{"target":"result-fusion","type":"related"},{"target":"score-fusion","type":"related"}],"representations":{"html":"https://id.searchplex.net/rank-fusion/","json":"https://id.searchplex.net/rank-fusion.json","jsonld":"https://id.searchplex.net/rank-fusion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Rank fusion does not necessarily require comparable raw scores across systems.","status":"published","uri":"https://id.searchplex.net/rank-fusion/"}
{"altLabels":["Rank-Aware Metric","Ranking Metric"],"definition":"A family of evaluation metrics whose value depends on where relevant items appear in the ranked result list.","derivedRelations":[{"target":"average-precision","type":"narrower"},{"target":"mean-average-precision","type":"narrower"},{"target":"mrr","type":"narrower"},{"target":"ndcg","type":"narrower"},{"target":"graded-relevance-metric","type":"related"}],"id":"rank-sensitive-metric","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Rank-Sensitive Metric","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"qrels","type":"related"},{"target":"ranking","type":"related"},{"target":"relevance-judgment","type":"related"}],"representations":{"html":"https://id.searchplex.net/rank-sensitive-metric/","json":"https://id.searchplex.net/rank-sensitive-metric.json","jsonld":"https://id.searchplex.net/rank-sensitive-metric.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Rank-sensitive metrics distinguish systems that place relevant results earlier from systems that retrieve the same relevant items lower in the ranking. Examples include MRR, Average Precision, MAP, and nDCG.","status":"published","uri":"https://id.searchplex.net/rank-sensitive-metric/"}
{"altLabels":["Relevance Ranking"],"definition":"Ordering retrieved or candidate items according to estimated relevance, utility, or another objective.","derivedRelations":[{"target":"average-precision","type":"evaluatedBy"},{"target":"beir","type":"evaluatedBy"},{"target":"mean-average-precision","type":"evaluatedBy"},{"target":"mrr","type":"evaluatedBy"},{"target":"ndcg","type":"evaluatedBy"},{"target":"precision-at-k","type":"evaluatedBy"},{"target":"success-at-k","type":"evaluatedBy"},{"target":"first-stage-ranking","type":"narrower"},{"target":"multi-stage-ranking","type":"narrower"},{"target":"reranking","type":"narrower"},{"note":"Boolean retrieval is based on satisfying logical conditions; ranking orders results by estimated relevance or utility.","target":"boolean-retrieval","type":"notEquivalentTo"},{"note":"Cross-Encoder names a model architecture or method; ranking is the task of ordering results.","target":"cross-encoder","type":"notEquivalentTo"},{"note":"Diversified retrieval adds an explicit coverage or diversity objective; ranking generally orders results by relevance or utility and may or may not optimize diversity.","target":"diversified-retrieval","type":"notEquivalentTo"},{"note":"Ranking is the task of ordering results; Learning to Rank is a method family for learning ranking functions.","target":"learning-to-rank","type":"notEquivalentTo"},{"note":"Ranking orders items; position bias is a behavioral effect of the displayed order.","target":"position-bias","type":"notEquivalentTo"},{"note":"Filtering restricts the eligible result set; ranking orders eligible results.","target":"result-filtering","type":"notEquivalentTo"},{"note":"Ranking orders results; a SERP is the page or interface where ranked and other search results are presented.","target":"serp","type":"notEquivalentTo"},{"target":"ad-hoc-retrieval","type":"related"},{"target":"beir","type":"related"},{"target":"cutoff-metric","type":"related"},{"target":"diversified-retrieval","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"federated-search","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"precision-metric","type":"related"},{"target":"query-classification","type":"related"},{"target":"rank-sensitive-metric","type":"related"},{"target":"retrieval","type":"related"},{"target":"search-as-you-type","type":"related"},{"target":"web-search","type":"related"},{"target":"bm25","type":"uses"},{"target":"bm25f","type":"uses"},{"target":"cross-encoder","type":"uses"},{"target":"learning-to-rank","type":"uses"},{"target":"maxsim","type":"uses"},{"target":"tf-idf","type":"uses"}],"distinctions":[{"reason":"Ranking is the ordering task; relevance is the criterion or relationship a ranking function often tries to estimate.","statement":"Ranking is not Relevance.","target":"relevance","targetLabel":"Relevance","targetUri":"https://id.searchplex.net/relevance/"}],"id":"ranking","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Scoring, term weighting and the vector space model","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/scoring-term-weighting-and-the-vector-space-model-1.html"}],"mappings":[{"relation":"closeMatch","scheme":"wikidata","uri":"https://www.wikidata.org/entity/Q163740"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Ranking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Ranking is the ordering task; relevance is the criterion or relationship a ranking function often tries to estimate.","target":"relevance","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"mrr","type":"related"},{"target":"ndcg","type":"related"},{"target":"popularity-bias","type":"related"},{"target":"position-bias","type":"related"},{"target":"relevance","type":"related"},{"target":"serp","type":"related"}],"representations":{"html":"https://id.searchplex.net/ranking/","json":"https://id.searchplex.net/ranking.json","jsonld":"https://id.searchplex.net/ranking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Ranking is about ordering; it is distinct from candidate generation, filtering, or pure matching.","status":"published","uri":"https://id.searchplex.net/ranking/"}
{"altLabels":["Recall-Oriented Metric","Recall-Based Metric"],"definition":"A family of retrieval or ranking evaluation metrics that emphasize coverage of relevant items.","derivedRelations":[{"target":"recall-at-k","type":"narrower"},{"target":"success-at-k","type":"narrower"}],"distinctions":[{"reason":"Recall metrics emphasize coverage of relevant items; precision metrics emphasize the fraction of returned results that are relevant.","statement":"Recall Metric is not Precision Metric.","target":"precision-metric","targetLabel":"Precision Metric","targetUri":"https://id.searchplex.net/precision-metric/"}],"id":"recall-metric","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Recall Metric","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Recall metrics emphasize coverage of relevant items; precision metrics emphasize the fraction of returned results that are relevant.","target":"precision-metric","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"cutoff-metric","type":"related"},{"target":"precision-metric","type":"related"}],"representations":{"html":"https://id.searchplex.net/recall-metric/","json":"https://id.searchplex.net/recall-metric.json","jsonld":"https://id.searchplex.net/recall-metric.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Recall-oriented metrics are especially important for candidate retrieval, high-recall retrieval, and first-stage retrieval, where the goal is to avoid losing relevant items before later ranking stages.","status":"published","uri":"https://id.searchplex.net/recall-metric/"}
{"definition":"An evaluation metric measuring the fraction of relevant items retrieved within the top K results.","derivedRelations":[{"note":"Success@K asks whether at least one relevant item appears in the top K; Recall@K measures how much of the relevant set appears in the top K.","target":"success-at-k","type":"notEquivalentTo"},{"target":"average-precision","type":"related"},{"target":"beir","type":"related"},{"target":"candidate-retrieval","type":"related"}],"distinctions":[{"reason":"Recall@K measures coverage of relevant items within the top K; Precision@K measures the fraction of top-K results that are relevant.","statement":"Recall@K is not Precision@K.","target":"precision-at-k","targetLabel":"Precision@K","targetUri":"https://id.searchplex.net/precision-at-k/"}],"id":"recall-at-k","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Evaluation of ranked retrieval results","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-ranked-retrieval-results-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Recall@K","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"cutoff-metric","type":"broader"},{"target":"recall-metric","type":"broader"},{"target":"candidate-retrieval","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"Recall@K measures coverage of relevant items within the top K; Precision@K measures the fraction of top-K results that are relevant.","target":"precision-at-k","type":"notEquivalentTo"},{"target":"ndcg","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"qrels","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"success-at-k","type":"related"}],"representations":{"html":"https://id.searchplex.net/recall-at-k/","json":"https://id.searchplex.net/recall-at-k.json","jsonld":"https://id.searchplex.net/recall-at-k.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Recall@K emphasizes coverage of relevant items at a cutoff and is often important when evaluating candidate retrieval or first-stage retrieval.","status":"published","uri":"https://id.searchplex.net/recall-at-k/"}
{"altLabels":["RRF"],"definition":"A rank-fusion method that combines multiple ranked result lists using reciprocal functions of item rank.","derivedRelations":[{"note":"Reciprocal Rank Fusion is a rank-fusion method, not a score-fusion method.","target":"score-fusion","type":"notEquivalentTo"},{"note":"RRF combines rank positions and avoids relying on comparable raw scores; score normalization transforms numerical scores.","target":"score-normalization","type":"notEquivalentTo"},{"target":"score-normalization","type":"related"}],"distinctions":[{"reason":"RRF can be used in hybrid retrieval systems, but hybrid retrieval is a broader retrieval pattern.","statement":"Reciprocal Rank Fusion is not Hybrid Retrieval.","target":"hybrid-retrieval","targetLabel":"Hybrid Retrieval","targetUri":"https://id.searchplex.net/hybrid-retrieval/"},{"reason":"RRF is one specific rank-fusion method, not the whole rank-fusion family.","statement":"Reciprocal Rank Fusion is not Rank Fusion.","target":"rank-fusion","targetLabel":"Rank Fusion","targetUri":"https://id.searchplex.net/rank-fusion/"}],"id":"reciprocal-rank-fusion","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Reciprocal Rank Fusion outperforms Condorcet and individual Rank Learning Methods","type":"definingReference","url":"https://doi.org/10.1145/1571941.1572114"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Reciprocal Rank Fusion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"rank-fusion","type":"broader"},{"note":"RRF can be used in hybrid retrieval systems, but hybrid retrieval is a broader retrieval pattern.","target":"hybrid-retrieval","type":"notEquivalentTo"},{"note":"RRF is one specific rank-fusion method, not the whole rank-fusion family.","target":"rank-fusion","type":"notEquivalentTo"},{"target":"hybrid-retrieval","type":"related"},{"target":"result-fusion","type":"related"}],"representations":{"html":"https://id.searchplex.net/reciprocal-rank-fusion/","json":"https://id.searchplex.net/reciprocal-rank-fusion.json","jsonld":"https://id.searchplex.net/reciprocal-rank-fusion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"RRF is one specific rank-fusion method. It should not be used as a synonym for rank fusion or hybrid retrieval generally.","slugAliases":["rrf"],"status":"published","uri":"https://id.searchplex.net/reciprocal-rank-fusion/"}
{"definition":"The degree to which a retrieved item satisfies an information need, query, topic, or evaluation criterion.","derivedRelations":[{"note":"LLM-as-judge may estimate or label relevance, but it is not the relevance relationship itself.","target":"llm-as-judge","type":"notEquivalentTo"},{"note":"Popularity bias favors popular items; relevance is the usefulness or topical fit of an item for a query or information need.","target":"popularity-bias","type":"notEquivalentTo"},{"note":"Position bias is caused by where an item is displayed; relevance is the item's usefulness or topical fit for the information need.","target":"position-bias","type":"notEquivalentTo"},{"target":"llm-as-judge","type":"related"},{"target":"popularity-bias","type":"related"},{"target":"position-bias","type":"related"},{"target":"qrels","type":"related"}],"distinctions":[{"reason":"Qrels are stored relevance judgments used for evaluation; relevance is the concept those judgments assess.","statement":"Relevance is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"Relevance is the relationship or criterion being estimated; ranking is the task of ordering results, often according to estimated relevance.","statement":"Relevance is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"Relevance is the underlying relationship between an item and an information need; a relevance judgment is an assessment or label of that relationship.","statement":"Relevance is not Relevance Judgment.","target":"relevance-judgment","targetLabel":"Relevance Judgment","targetUri":"https://id.searchplex.net/relevance-judgment/"}],"id":"relevance","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Relevance: The whole history","type":"canonicalSurvey","url":"https://doi.org/10.1002/(SICI)1097-4571(199701)48:9%3C810::AID-ASI6%3E3.0.CO;2-U"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Relevance","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Qrels are stored relevance judgments used for evaluation; relevance is the concept those judgments assess.","target":"qrels","type":"notEquivalentTo"},{"note":"Relevance is the relationship or criterion being estimated; ranking is the task of ordering results, often according to estimated relevance.","target":"ranking","type":"notEquivalentTo"},{"note":"Relevance is the underlying relationship between an item and an information need; a relevance judgment is an assessment or label of that relationship.","target":"relevance-judgment","type":"notEquivalentTo"},{"target":"ranking","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/relevance/","json":"https://id.searchplex.net/relevance.json","jsonld":"https://id.searchplex.net/relevance.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In information retrieval evaluation, relevance may be binary or graded and is usually operationalized through relevance judgments. Relevance is the evaluation target, not itself a metric or a qrels file.","status":"published","uri":"https://id.searchplex.net/relevance/"}
{"altLabels":["Relevance Assessment","Relevance Judgments","Relevance Judgements"],"definition":"An assessment assigning a relevance label or level to an item with respect to a query, topic, or information need.","derivedRelations":[{"note":"Pooling identifies candidates for assessment; a relevance judgment is an assessment of an item for a topic or query.","target":"pooling","type":"notEquivalentTo"},{"note":"Relevance is the underlying relationship between an item and an information need; a relevance judgment is an assessment or label of that relationship.","target":"relevance","type":"notEquivalentTo"},{"target":"graded-relevance-metric","type":"related"},{"target":"information-filtering","type":"related"},{"target":"learning-to-rank","type":"related"},{"target":"mrr","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"rank-sensitive-metric","type":"related"}],"distinctions":[{"reason":"A relevance judgment is an assessment; LLM-as-judge is a method that may produce or assist such assessments.","statement":"Relevance Judgment is not LLM-as-Judge.","target":"llm-as-judge","targetLabel":"LLM-as-Judge","targetUri":"https://id.searchplex.net/llm-as-judge/"},{"reason":"Relevance judgments provide labels used by metrics such as nDCG; nDCG is an evaluation metric, not a judgment.","statement":"Relevance Judgment is not nDCG@K.","target":"ndcg","targetLabel":"nDCG@K","targetUri":"https://id.searchplex.net/ndcg/"},{"reason":"A relevance judgment is an individual assessment; qrels are a structured set or file of judgments used in retrieval evaluation.","statement":"Relevance Judgment is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"}],"id":"relevance-judgment","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Text REtrieval Conference","type":"officialHomepage","url":"https://trec.nist.gov/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Relevance Judgment","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"A relevance judgment is an assessment; LLM-as-judge is a method that may produce or assist such assessments.","target":"llm-as-judge","type":"notEquivalentTo"},{"note":"Relevance judgments provide labels used by metrics such as nDCG; nDCG is an evaluation metric, not a judgment.","target":"ndcg","type":"notEquivalentTo"},{"note":"A relevance judgment is an individual assessment; qrels are a structured set or file of judgments used in retrieval evaluation.","target":"qrels","type":"notEquivalentTo"},{"target":"llm-as-judge","type":"related"},{"target":"ndcg","type":"related"},{"target":"pooling","type":"related"},{"target":"qrels","type":"related"},{"target":"recall-at-k","type":"related"},{"target":"relevance","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/relevance-judgment/","json":"https://id.searchplex.net/relevance-judgment.json","jsonld":"https://id.searchplex.net/relevance-judgment.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Relevance judgments are commonly produced by human assessors in Cranfield-style and TREC-style evaluation, though automatic or synthetic judgments may also be studied. Judgments may be binary or graded and are often collected into qrels.","status":"published","uri":"https://id.searchplex.net/relevance-judgment/"}
{"altLabels":["Re-ranking"],"definition":"Re-scoring or reordering a bounded candidate set using a subsequent ranking model or additional evidence.","derivedRelations":[{"target":"mrr","type":"evaluatedBy"},{"target":"mteb","type":"evaluatedBy"},{"target":"ndcg","type":"evaluatedBy"},{"note":"A Cross-Encoder is one common way to perform reranking, but reranking is the retrieval-system task or stage.","target":"cross-encoder","type":"notEquivalentTo"},{"note":"Reranking is a pipeline stage or task; Learning to Rank methods may be used for reranking but are not the stage itself.","target":"learning-to-rank","type":"notEquivalentTo"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"mteb","type":"related"},{"target":"cross-encoder","type":"uses"},{"target":"learning-to-rank","type":"uses"}],"id":"reranking","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Retrieve & Re-Rank","type":"officialHomepage","url":"https://www.sbert.net/examples/applications/retrieve_rerank/README.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Reranking","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"ranking","type":"broader"},{"target":"multi-stage-ranking","type":"partOf"},{"target":"retrieve-and-rerank","type":"partOf"},{"target":"candidate-retrieval","type":"related"},{"target":"first-stage-ranking","type":"related"}],"representations":{"html":"https://id.searchplex.net/reranking/","json":"https://id.searchplex.net/reranking.json","jsonld":"https://id.searchplex.net/reranking.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Reranking normally operates over candidates already selected by an earlier retrieval or ranking stage and can therefore use more expensive models.","status":"published","terms":[{"community":"industry","label":"Second-Stage Ranking","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/reranking/"}
{"altLabels":["Search Filtering"],"definition":"A method for restricting search or retrieval results to items that satisfy explicit constraints, predicates, or selected attribute values.","derivedRelations":[{"note":"Routing is a task formulation over incoming documents; result filtering constrains or removes results using predicates or facets.","target":"routing","type":"notEquivalentTo"}],"distinctions":[{"reason":"Result filtering is the constraint operation; faceted search is a structured navigation model that often uses filters.","statement":"Result Filtering is not Faceted Search.","target":"faceted-search","targetLabel":"Faceted Search","targetUri":"https://id.searchplex.net/faceted-search/"},{"reason":"Result filtering applies constraints to search results or eligible items; information filtering is a task involving stable needs or profiles over incoming information.","statement":"Result Filtering is not Information Filtering.","target":"information-filtering","targetLabel":"Information Filtering","targetUri":"https://id.searchplex.net/information-filtering/"},{"reason":"Result filtering restricts eligible results; query relaxation broadens eligibility or weakens matching criteria.","statement":"Result Filtering is not Query Relaxation.","target":"query-relaxation","targetLabel":"Query Relaxation","targetUri":"https://id.searchplex.net/query-relaxation/"},{"reason":"Filtering restricts the eligible result set; ranking orders eligible results.","statement":"Result Filtering is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"}],"id":"result-filtering","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Filters vs. Facets","type":"canonicalSurvey","url":"https://www.nngroup.com/articles/filters-vs-facets/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Result Filtering","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Result filtering is the constraint operation; faceted search is a structured navigation model that often uses filters.","target":"faceted-search","type":"notEquivalentTo"},{"note":"Result filtering applies constraints to search results or eligible items; information filtering is a task involving stable needs or profiles over incoming information.","target":"information-filtering","type":"notEquivalentTo"},{"note":"Result filtering restricts eligible results; query relaxation broadens eligibility or weakens matching criteria.","target":"query-relaxation","type":"notEquivalentTo"},{"note":"Filtering restricts the eligible result set; ranking orders eligible results.","target":"ranking","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"faceted-search","type":"related"},{"target":"query-relaxation","type":"related"}],"representations":{"html":"https://id.searchplex.net/result-filtering/","json":"https://id.searchplex.net/result-filtering.json","jsonld":"https://id.searchplex.net/result-filtering.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Result filtering may be applied before retrieval, during retrieval, after retrieval, or through a user interface. It is distinct from faceted search, which presents structured dimensions for exploratory narrowing, and from information filtering, which is a standing-query task over incoming information.","status":"published","uri":"https://id.searchplex.net/result-filtering/"}
{"definition":"Combining outputs from multiple retrieval or ranking processes into a single result set or ranking.","derivedRelations":[{"note":"Result fusion combines ranked lists or result sets; diversified retrieval is an objective about covering aspects or intents in the returned results.","target":"diversified-retrieval","type":"notEquivalentTo"},{"note":"Result fusion is a combining operation; federated search is the broader multi-source search task or system pattern that often uses result fusion.","target":"federated-search","type":"notEquivalentTo"}],"id":"result-fusion","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An Analysis of Fusion Functions for Hybrid Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2210.11934"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Result Fusion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"federated-search","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"rank-fusion","type":"related"},{"target":"reciprocal-rank-fusion","type":"related"},{"target":"score-fusion","type":"related"}],"representations":{"html":"https://id.searchplex.net/result-fusion/","json":"https://id.searchplex.net/result-fusion.json","jsonld":"https://id.searchplex.net/result-fusion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Result fusion is an umbrella concept that may operate on ranks, scores, probabilities, or other evidence.","status":"published","uri":"https://id.searchplex.net/result-fusion/"}
{"definition":"Selecting items from a collection that are estimated to be relevant or useful for a query or information need.","derivedRelations":[{"target":"average-precision","type":"evaluatedBy"},{"target":"beir","type":"evaluatedBy"},{"target":"mean-average-precision","type":"evaluatedBy"},{"target":"mteb","type":"evaluatedBy"},{"target":"precision-at-k","type":"evaluatedBy"},{"target":"recall-at-k","type":"evaluatedBy"},{"target":"success-at-k","type":"evaluatedBy"},{"target":"ad-hoc-retrieval","type":"narrower"},{"target":"agentic-retrieval","type":"narrower"},{"target":"diversified-retrieval","type":"narrower"},{"target":"document-retrieval","type":"narrower"},{"target":"entity-search","type":"narrower"},{"target":"federated-search","type":"narrower"},{"target":"information-filtering","type":"narrower"},{"target":"interactive-search","type":"narrower"},{"target":"iterative-retrieval","type":"narrower"},{"target":"known-item-search","type":"narrower"},{"target":"multimodal-retrieval","type":"narrower"},{"target":"passage-retrieval","type":"narrower"},{"target":"question-answering","type":"narrower"},{"target":"routing","type":"narrower"},{"target":"single-shot-retrieval","type":"narrower"},{"target":"web-search","type":"narrower"},{"note":"Chunking creates or selects retrievable units; retrieval selects relevant units from an indexed collection.","target":"chunking","type":"notEquivalentTo"},{"note":"Retrieval selects items; a SERP presents retrieved results and related search features to the user.","target":"serp","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"chunking","type":"related"},{"target":"cross-language-retrieval","type":"related"},{"target":"cutoff-metric","type":"related"},{"target":"document-expansion","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"mteb","type":"related"},{"target":"multilingual-retrieval","type":"related"},{"target":"query-processing","type":"related"},{"target":"relevance","type":"related"},{"target":"serp","type":"related"},{"target":"test-collection","type":"related"},{"target":"approximate-nearest-neighbor-search","type":"uses"},{"target":"bm25","type":"uses"},{"target":"bm25f","type":"uses"},{"target":"boolean-retrieval","type":"uses"},{"target":"hnsw","type":"uses"},{"target":"inverted-index","type":"uses"},{"target":"tf-idf","type":"uses"}],"distinctions":[{"reason":"Retrieval selects items for a query; agentic retrieval describes an agent-controlled loop of retrieval actions.","statement":"Retrieval is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Retrieval can supply source material for grounding, but grounding concerns whether output is supported by source material.","statement":"Retrieval is not Grounding.","target":"grounding","targetLabel":"Grounding","targetUri":"https://id.searchplex.net/grounding/"},{"reason":"Retrieval selects relevant items; RAG is a generation pattern that uses retrieved information.","statement":"Retrieval is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"}],"id":"retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An example information retrieval problem","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Retrieval selects items for a query; agentic retrieval describes an agent-controlled loop of retrieval actions.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Retrieval can supply source material for grounding, but grounding concerns whether output is supported by source material.","target":"grounding","type":"notEquivalentTo"},{"note":"Retrieval selects relevant items; RAG is a generation pattern that uses retrieved information.","target":"rag","type":"notEquivalentTo"},{"target":"candidate-retrieval","type":"related"},{"target":"grounding","type":"related"},{"target":"multi-stage-ranking","type":"related"},{"target":"rag","type":"related"},{"target":"ranking","type":"related"}],"representations":{"html":"https://id.searchplex.net/retrieval/","json":"https://id.searchplex.net/retrieval.json","jsonld":"https://id.searchplex.net/retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Retrieval may itself produce scores or an initial ordering. In multi-stage systems, it is often distinguished from more expensive downstream ranking or reranking.","status":"published","uri":"https://id.searchplex.net/retrieval/"}
{"definition":"Choosing a sequence or strategy of retrieval actions for satisfying an information need, often before or during an agentic retrieval loop.","derivedRelations":[{"note":"Retrieval planning is one capability within an agentic RAG system; it is not the full RAG pattern.","target":"agentic-rag","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"deep-research","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"web-browsing-agent","type":"related"}],"distinctions":[{"reason":"Retrieval planning is one capability within agentic retrieval; agentic retrieval is the broader agent-controlled retrieval pattern.","statement":"Retrieval Planning is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Retrieval planning may use query decomposition, but decomposition specifically breaks an information need into subqueries or subproblems.","statement":"Retrieval Planning is not Query Decomposition.","target":"query-decomposition","targetLabel":"Query Decomposition","targetUri":"https://id.searchplex.net/query-decomposition/"},{"reason":"Retrieval planning chooses an overall retrieval strategy or sequence; query routing chooses a handling path, source, retriever, tool, or model for a query.","statement":"Retrieval Planning is not Query Routing.","target":"query-routing","targetLabel":"Query Routing","targetUri":"https://id.searchplex.net/query-routing/"}],"id":"retrieval-planning","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Agentic Retrieval-Augmented Generation","type":"canonicalSurvey","url":"https://arxiv.org/abs/2501.09136"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Retrieval Planning","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Retrieval planning is one capability within agentic retrieval; agentic retrieval is the broader agent-controlled retrieval pattern.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Retrieval planning may use query decomposition, but decomposition specifically breaks an information need into subqueries or subproblems.","target":"query-decomposition","type":"notEquivalentTo"},{"note":"Retrieval planning chooses an overall retrieval strategy or sequence; query routing chooses a handling path, source, retriever, tool, or model for a query.","target":"query-routing","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"partOf"},{"target":"multi-hop-retrieval","type":"related"},{"target":"query-decomposition","type":"related"},{"target":"query-routing","type":"related"},{"target":"rag","type":"related"}],"representations":{"html":"https://id.searchplex.net/retrieval-planning/","json":"https://id.searchplex.net/retrieval-planning.json","jsonld":"https://id.searchplex.net/retrieval-planning.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Retrieval planning may include deciding whether retrieval is needed, decomposing an information need, choosing tools or indexes, ordering retrieval steps, and deciding when to stop. It is distinct from query routing, which chooses where or how a query should be handled, and from query decomposition, which breaks an information need into subqueries.","status":"published","terms":[{"community":"industry","label":"Search Planning","usage":"nearSynonym"},{"community":"industry","label":"Retrieval Strategy Planning","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/retrieval-planning/"}
{"altLabels":["RAG"],"definition":"A generation pattern in which retrieved external information is supplied to a generative model to support, condition, or ground its output.","derivedRelations":[{"note":"Chunking is one design choice inside many RAG systems; RAG is the broader retrieval-augmented generation pattern.","target":"chunking","type":"notEquivalentTo"},{"note":"Conversational search is a multi-turn search interaction pattern; RAG is a generation pattern using retrieved information.","target":"conversational-search","type":"notEquivalentTo"},{"note":"RAG is a retrieval-augmented generation pattern; deep research emphasizes multi-step search, browsing, evidence gathering, and cited synthesis.","target":"deep-research","type":"notEquivalentTo"},{"note":"Multimodal retrieval may be used inside multimodal RAG systems, but RAG is a retrieval-augmented generation pattern.","target":"multimodal-retrieval","type":"notEquivalentTo"},{"note":"Passage retrieval is often used inside RAG systems, but RAG is the broader retrieval-augmented generation pattern.","target":"passage-retrieval","type":"notEquivalentTo"},{"note":"Question answering is a task; RAG is a retrieval-augmented generation pattern that can be used for question answering and other tasks.","target":"question-answering","type":"notEquivalentTo"},{"target":"conversational-search","type":"related"},{"target":"deep-research","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"multimodal-retrieval","type":"related"},{"target":"query-routing","type":"related"},{"target":"question-answering","type":"related"},{"target":"retrieval-planning","type":"related"}],"distinctions":[{"reason":"RAG is the broader retrieval-augmented generation pattern; Agentic RAG specifically adds agentic control over retrieval or tool-use steps.","statement":"Retrieval-Augmented Generation is not Agentic RAG.","target":"agentic-rag","targetLabel":"Agentic RAG","targetUri":"https://id.searchplex.net/agentic-rag/"},{"reason":"RAG is a generation pattern using retrieved information; agentic retrieval is an agent-controlled retrieval pattern that may be used within RAG systems.","statement":"Retrieval-Augmented Generation is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Dense retrieval is one possible retrieval component in RAG; RAG is the broader system pattern.","statement":"Retrieval-Augmented Generation is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"RAG is one way to provide external context for grounding; grounding is the broader goal or practice of anchoring output in source material.","statement":"Retrieval-Augmented Generation is not Grounding.","target":"grounding","targetLabel":"Grounding","targetUri":"https://id.searchplex.net/grounding/"},{"reason":"Retrieval is the selection of relevant items; RAG is a generation pattern that uses retrieved information as input to a generative model.","statement":"Retrieval-Augmented Generation is not Retrieval.","target":"retrieval","targetLabel":"Retrieval","targetUri":"https://id.searchplex.net/retrieval/"},{"reason":"Semantic search is meaning-oriented retrieval; RAG is a retrieval-augmented generation pattern that may use semantic search.","statement":"Retrieval-Augmented Generation is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"}],"id":"rag","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks","type":"definingReference","url":"https://arxiv.org/abs/2005.11401"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Retrieval-Augmented Generation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"RAG is the broader retrieval-augmented generation pattern; Agentic RAG specifically adds agentic control over retrieval or tool-use steps.","target":"agentic-rag","type":"notEquivalentTo"},{"note":"RAG is a generation pattern using retrieved information; agentic retrieval is an agent-controlled retrieval pattern that may be used within RAG systems.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Dense retrieval is one possible retrieval component in RAG; RAG is the broader system pattern.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"RAG is one way to provide external context for grounding; grounding is the broader goal or practice of anchoring output in source material.","target":"grounding","type":"notEquivalentTo"},{"note":"Retrieval is the selection of relevant items; RAG is a generation pattern that uses retrieved information as input to a generative model.","target":"retrieval","type":"notEquivalentTo"},{"note":"Semantic search is meaning-oriented retrieval; RAG is a retrieval-augmented generation pattern that may use semantic search.","target":"semantic-search","type":"notEquivalentTo"},{"target":"agentic-rag","type":"related"},{"target":"agentic-retrieval","type":"related"},{"target":"chunking","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"grounding","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"passage-retrieval","type":"related"},{"target":"retrieval","type":"related"},{"target":"semantic-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/rag/","json":"https://id.searchplex.net/rag.json","jsonld":"https://id.searchplex.net/rag.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"RAG systems commonly combine indexing, retrieval, reranking, prompt or context construction, and generation. RAG uses retrieval, but it is not identical to retrieval, dense retrieval, semantic search, or grounding.","status":"published","uri":"https://id.searchplex.net/rag/"}
{"altLabels":["Retrieve-Then-Rerank"],"definition":"A multi-stage search pattern in which an efficient retrieval stage first produces a bounded candidate set and a subsequent, typically more expensive model rescoring or reorders those candidates.","derivedRelations":[{"target":"candidate-retrieval","type":"hasPart"},{"target":"reranking","type":"hasPart"},{"note":"Retrieve-and-rerank retrieves a candidate set and then reranks it; multi-hop retrieval performs multiple dependent retrieval steps.","target":"multi-hop-retrieval","type":"notEquivalentTo"}],"distinctions":[{"reason":"Retrieve-and-rerank is a pipeline pattern; a Cross-Encoder is one possible reranking method.","statement":"Retrieve-and-Rerank is not Cross-Encoder.","target":"cross-encoder","targetLabel":"Cross-Encoder","targetUri":"https://id.searchplex.net/cross-encoder/"},{"reason":"Retrieve-and-rerank is a pipeline pattern; a dual encoder is one possible first-stage method.","statement":"Retrieve-and-Rerank is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"}],"id":"retrieve-and-rerank","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Sentence Transformers Retrieve & Re-Rank","type":"officialHomepage","url":"https://www.sbert.net/examples/applications/retrieve_rerank/README.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Retrieve-and-Rerank","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"multi-stage-ranking","type":"broader"},{"note":"Retrieve-and-rerank is a pipeline pattern; a Cross-Encoder is one possible reranking method.","target":"cross-encoder","type":"notEquivalentTo"},{"note":"Retrieve-and-rerank is a pipeline pattern; a dual encoder is one possible first-stage method.","target":"dual-encoder","type":"notEquivalentTo"},{"target":"cross-encoder","type":"related"},{"target":"dual-encoder","type":"related"}],"representations":{"html":"https://id.searchplex.net/retrieve-and-rerank/","json":"https://id.searchplex.net/retrieve-and-rerank.json","jsonld":"https://id.searchplex.net/retrieve-and-rerank.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"A common neural implementation uses a bi- or dual-encoder for candidate retrieval followed by a Cross-Encoder for reranking, but retrieve-and-rerank does not require those particular model architectures. The first stage may instead be lexical, sparse, hybrid, or another retrieval method.","status":"published","uri":"https://id.searchplex.net/retrieve-and-rerank/"}
{"altLabels":["Reverse Image Retrieval","Search by Image","Image-to-Image Search","Similar Image Search"],"definition":"An image-retrieval task in which the user provides an image, image URL, or visual example as the query to find visually similar, related, source, duplicate, or derivative images.","distinctions":[{"reason":"Reverse image search is usually image-to-image or image-query retrieval; cross-modal retrieval crosses between different modalities such as text and image.","statement":"Reverse Image Search is not Cross-Modal Retrieval.","target":"cross-modal-retrieval","targetLabel":"Cross-Modal Retrieval","targetUri":"https://id.searchplex.net/cross-modal-retrieval/"},{"reason":"Reverse image search is a specific form of image retrieval where the query is an image or visual example.","statement":"Reverse Image Search is not Image Retrieval.","target":"image-retrieval","targetLabel":"Image Retrieval","targetUri":"https://id.searchplex.net/image-retrieval/"},{"reason":"Reverse image search may use vector search internally, but it is a user-facing image-query retrieval task.","statement":"Reverse Image Search is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"reverse-image-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"A Decade Survey of Content Based Image Retrieval using Deep Learning","type":"canonicalSurvey","url":"https://arxiv.org/abs/2012.00641"},{"label":"Advancements in Content-Based Image Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2312.10089"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Reverse Image Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"image-retrieval","type":"broader"},{"note":"Reverse image search is usually image-to-image or image-query retrieval; cross-modal retrieval crosses between different modalities such as text and image.","target":"cross-modal-retrieval","type":"notEquivalentTo"},{"note":"Reverse image search is a specific form of image retrieval where the query is an image or visual example.","target":"image-retrieval","type":"notEquivalentTo"},{"note":"Reverse image search may use vector search internally, but it is a user-facing image-query retrieval task.","target":"vector-search","type":"notEquivalentTo"},{"target":"approximate-nearest-neighbor-search","type":"related"},{"target":"embedding","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/reverse-image-search/","json":"https://id.searchplex.net/reverse-image-search.json","jsonld":"https://id.searchplex.net/reverse-image-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Reverse image search is often implemented with content-based image retrieval techniques, visual embeddings, nearest-neighbor search, or perceptual similarity methods. It is distinct from image retrieval in general and from text-to-image cross-modal retrieval.","status":"published","terms":[{"community":"academic","label":"Content-Based Image Retrieval","usage":"nearSynonym"},{"community":"academic","label":"CBIR","usage":"abbreviation"},{"community":"industry","label":"Query by Image","usage":"nearSynonym"},{"community":"industry","label":"Image Query Search","usage":"informal"},{"community":"academic","label":"Query by Example","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/reverse-image-search/"}
{"altLabels":["Information Routing"],"definition":"An IR task in which standing queries or profiles are used to retrieve relevant items from a stream or changing supply of new documents.","distinctions":[{"reason":"Routing applies persistent information needs to new documents; ad hoc retrieval applies new information needs to a known collection.","statement":"Routing is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Routing in classic IR uses standing queries or profiles over incoming documents; query routing chooses where or how a live query should be handled.","statement":"Routing is not Query Routing.","target":"query-routing","targetLabel":"Query Routing","targetUri":"https://id.searchplex.net/query-routing/"},{"reason":"Routing is a task formulation over incoming documents; result filtering constrains or removes results using predicates or facets.","statement":"Routing is not Result Filtering.","target":"result-filtering","targetLabel":"Result Filtering","targetUri":"https://id.searchplex.net/result-filtering/"}],"id":"routing","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"The Text Retrieval Conferences (TRECs)","type":"definingReference","url":"https://aclanthology.org/X98-1027/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Routing","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Routing applies persistent information needs to new documents; ad hoc retrieval applies new information needs to a known collection.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Routing in classic IR uses standing queries or profiles over incoming documents; query routing chooses where or how a live query should be handled.","target":"query-routing","type":"notEquivalentTo"},{"note":"Routing is a task formulation over incoming documents; result filtering constrains or removes results using predicates or facets.","target":"result-filtering","type":"notEquivalentTo"},{"target":"ad-hoc-retrieval","type":"related"},{"target":"information-filtering","type":"related"}],"representations":{"html":"https://id.searchplex.net/routing/","json":"https://id.searchplex.net/routing.json","jsonld":"https://id.searchplex.net/routing.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Routing is a classic TREC task formulation associated with standing information needs and incoming documents. It is distinct from ad hoc retrieval, where new topics are run against a known collection, and from query routing, where a live query is directed to a source, model, retriever, or handling strategy.","status":"published","terms":[{"community":"academic","label":"Selective Dissemination of Information","usage":"historical"},{"community":"academic","label":"SDI","usage":"abbreviation"}],"uri":"https://id.searchplex.net/routing/"}
{"artifactKind":"Dataset","definition":"A scientific claim verification and evidence-retrieval dataset used to evaluate retrieval and fact-verification systems.","distinctions":[{"reason":"SciFact is a component evaluation dataset; BEIR is the benchmark suite that includes SciFact among other datasets.","statement":"SciFact is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"SciFact includes relevance or evidence judgments; qrels are judgment files, not the dataset identity.","statement":"SciFact is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"}],"id":"scifact","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"SciFact repository","type":"officialHomepage","url":"https://github.com/allenai/scifact"},{"label":"SciFact on Hugging Face","type":"officialHomepage","url":"https://huggingface.co/datasets/BeIR/scifact"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"SciFact","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"SciFact is a component evaluation dataset; BEIR is the benchmark suite that includes SciFact among other datasets.","target":"beir","type":"notEquivalentTo"},{"note":"SciFact includes relevance or evidence judgments; qrels are judgment files, not the dataset identity.","target":"qrels","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/scifact/","json":"https://id.searchplex.net/scifact.json","jsonld":"https://id.searchplex.net/scifact.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"SciFact is commonly used as a BEIR component dataset for scientific fact-checking retrieval. It is an evaluation dataset, not the BEIR benchmark suite itself.","status":"published","uri":"https://id.searchplex.net/scifact/"}
{"definition":"A family of fusion methods that combine numerical scores produced by multiple retrieval or ranking systems.","distinctions":[{"reason":"Score fusion combines numerical scores; rank fusion combines rank positions or rank-derived evidence.","statement":"Score Fusion is not Rank Fusion.","target":"rank-fusion","targetLabel":"Rank Fusion","targetUri":"https://id.searchplex.net/rank-fusion/"},{"reason":"Reciprocal Rank Fusion is a rank-fusion method, not a score-fusion method.","statement":"Score Fusion is not Reciprocal Rank Fusion.","target":"reciprocal-rank-fusion","targetLabel":"Reciprocal Rank Fusion","targetUri":"https://id.searchplex.net/reciprocal-rank-fusion/"},{"reason":"Score normalization prepares scores for comparison or combination; score fusion combines scores from multiple systems.","statement":"Score Fusion is not Score Normalization.","target":"score-normalization","targetLabel":"Score Normalization","targetUri":"https://id.searchplex.net/score-normalization/"}],"id":"score-fusion","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An Analysis of Fusion Functions for Hybrid Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2210.11934"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Score Fusion","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Score fusion combines numerical scores; rank fusion combines rank positions or rank-derived evidence.","target":"rank-fusion","type":"notEquivalentTo"},{"note":"Reciprocal Rank Fusion is a rank-fusion method, not a score-fusion method.","target":"reciprocal-rank-fusion","type":"notEquivalentTo"},{"note":"Score normalization prepares scores for comparison or combination; score fusion combines scores from multiple systems.","target":"score-normalization","type":"notEquivalentTo"},{"target":"hybrid-retrieval","type":"related"},{"target":"rank-fusion","type":"related"},{"target":"result-fusion","type":"related"},{"target":"score-normalization","type":"related"}],"representations":{"html":"https://id.searchplex.net/score-fusion/","json":"https://id.searchplex.net/score-fusion.json","jsonld":"https://id.searchplex.net/score-fusion.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Score fusion commonly requires normalization, calibration, weighting, or another way to make scores from different systems meaningfully combinable.","status":"published","uri":"https://id.searchplex.net/score-fusion/"}
{"altLabels":["Score Normalisation","Score Scaling"],"definition":"A method for transforming scores from one or more retrieval or ranking systems onto a scale intended to make them comparable or combinable.","distinctions":[{"reason":"Score normalization may be used inside a hybrid retrieval system, but hybrid retrieval is the broader combination of retrieval signals.","statement":"Score Normalization is not Hybrid Retrieval.","target":"hybrid-retrieval","targetLabel":"Hybrid Retrieval","targetUri":"https://id.searchplex.net/hybrid-retrieval/"},{"reason":"RRF combines rank positions and avoids relying on comparable raw scores; score normalization transforms numerical scores.","statement":"Score Normalization is not Reciprocal Rank Fusion.","target":"reciprocal-rank-fusion","targetLabel":"Reciprocal Rank Fusion","targetUri":"https://id.searchplex.net/reciprocal-rank-fusion/"},{"reason":"Score normalization prepares scores for comparison or combination; score fusion combines the normalized or calibrated scores.","statement":"Score Normalization is not Score Fusion.","target":"score-fusion","targetLabel":"Score Fusion","targetUri":"https://id.searchplex.net/score-fusion/"}],"id":"score-normalization","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An Analysis of Fusion Functions for Hybrid Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/2210.11934"},{"label":"Hybrid search depends on query type","type":"officialHomepage","url":"https://hornet.dev/blog/100m-doc-search-part-3-hybrid-search"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Score Normalization","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Score normalization may be used inside a hybrid retrieval system, but hybrid retrieval is the broader combination of retrieval signals.","target":"hybrid-retrieval","type":"notEquivalentTo"},{"note":"RRF combines rank positions and avoids relying on comparable raw scores; score normalization transforms numerical scores.","target":"reciprocal-rank-fusion","type":"notEquivalentTo"},{"note":"Score normalization prepares scores for comparison or combination; score fusion combines the normalized or calibrated scores.","target":"score-fusion","type":"notEquivalentTo"},{"target":"hybrid-retrieval","type":"related"},{"target":"rank-fusion","type":"related"},{"target":"reciprocal-rank-fusion","type":"related"},{"target":"score-fusion","type":"related"}],"representations":{"html":"https://id.searchplex.net/score-normalization/","json":"https://id.searchplex.net/score-normalization.json","jsonld":"https://id.searchplex.net/score-normalization.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Score normalization is commonly used before weighted score fusion in hybrid retrieval. It is a score-preparation step, not the fusion method itself, and it is distinct from rank-based methods such as Reciprocal Rank Fusion.","status":"published","terms":[{"community":"industry","label":"Linear Normalization","usage":"nearSynonym"},{"community":"academic","label":"Min-Max Normalization","usage":"nearSynonym"},{"community":"academic","label":"Z-Score Normalization","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/score-normalization/"}
{"altLabels":["Search As You Type","Typeahead Search"],"definition":"A search interaction pattern in which retrieval is executed and results are updated continuously as the user types a query.","distinctions":[{"reason":"Search-as-you-type updates actual search results while the query is being entered; query autocomplete proposes completions for the query prefix.","statement":"Search-as-You-Type is not Query Autocomplete.","target":"query-autocomplete","targetLabel":"Query Autocomplete","targetUri":"https://id.searchplex.net/query-autocomplete/"},{"reason":"Search-as-you-type runs retrieval during typing; query suggestion proposes related, next, or alternative queries.","statement":"Search-as-You-Type is not Query Suggestion.","target":"query-suggestion","targetLabel":"Query Suggestion","targetUri":"https://id.searchplex.net/query-suggestion/"},{"reason":"Search-as-you-type is an interaction pattern; query understanding is the interpretation of the query.","statement":"Search-as-You-Type is not Query Understanding.","target":"query-understanding","targetLabel":"Query Understanding","targetUri":"https://id.searchplex.net/query-understanding/"}],"id":"search-as-you-type","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Elasticsearch autocomplete: search as you type","type":"officialHomepage","url":"https://www.elastic.co/search-labs/blog/elasticsearch-autocomplete-search"},{"label":"Algolia Autocomplete","type":"officialHomepage","url":"https://www.algolia.com/doc/guides/building-search-ui/ui-and-ux-patterns/autocomplete/js"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Search-as-You-Type","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Search-as-you-type updates actual search results while the query is being entered; query autocomplete proposes completions for the query prefix.","target":"query-autocomplete","type":"notEquivalentTo"},{"note":"Search-as-you-type runs retrieval during typing; query suggestion proposes related, next, or alternative queries.","target":"query-suggestion","type":"notEquivalentTo"},{"note":"Search-as-you-type is an interaction pattern; query understanding is the interpretation of the query.","target":"query-understanding","type":"notEquivalentTo"},{"target":"query-autocomplete","type":"related"},{"target":"query-suggestion","type":"related"},{"target":"query-understanding","type":"related"},{"target":"ranking","type":"related"},{"target":"spelling-correction","type":"related"}],"representations":{"html":"https://id.searchplex.net/search-as-you-type/","json":"https://id.searchplex.net/search-as-you-type.json","jsonld":"https://id.searchplex.net/search-as-you-type.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Search-as-you-type returns or updates actual search results during query entry. It is often combined with autocomplete and query suggestions, but it is distinct from query autocomplete, which proposes query completions, and from query suggestion, which recommends related or alternative queries.","status":"published","terms":[{"community":"industry","label":"Instant Search","usage":"nearSynonym"},{"community":"industry","label":"Live Search","usage":"nearSynonym"},{"community":"industry","label":"As-You-Type Search","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/search-as-you-type/"}
{"definition":"A broad term for search intended to retrieve according to meaning or semantic relatedness rather than relying only on literal lexical overlap.","derivedRelations":[{"note":"Semantic search describes meaning-oriented retrieval; agentic retrieval describes who controls and iterates the retrieval process.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Semantic search is a broad meaning-oriented search goal or product capability; Dense Passage Retrieval is one method that can support such systems.","target":"dense-passage-retrieval","type":"notEquivalentTo"},{"note":"Semantic search is a broad meaning-oriented search goal; entity search is a task centered on entities.","target":"entity-search","type":"notEquivalentTo"},{"note":"Exact match requires literal matching; semantic search is meaning-oriented and may retrieve results without exact lexical overlap.","target":"exact-match","type":"notEquivalentTo"},{"note":"Image retrieval names the target modality; semantic search names a meaning-oriented retrieval goal or capability.","target":"image-retrieval","type":"notEquivalentTo"},{"note":"Semantic search names a meaning-oriented retrieval goal or product capability; single-vector retrieval names one possible representation format.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"note":"Web search is a search setting and task family; semantic search is a broad approach or capability focused on meaning-based matching.","target":"web-search","type":"notEquivalentTo"},{"target":"dense-passage-retrieval","type":"related"},{"target":"entity-search","type":"related"}],"distinctions":[{"reason":"Semantic search may use dense retrieval, but it is a broader product or intent concept rather than a specific representation family.","statement":"Semantic Search is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Semantic search retrieves by meaning; grounding concerns whether output is supported by source material.","statement":"Semantic Search is not Grounding.","target":"grounding","targetLabel":"Grounding","targetUri":"https://id.searchplex.net/grounding/"},{"reason":"Semantic search describes meaning-oriented retrieval; keyword search describes retrieval centered on explicit query keywords or phrases.","statement":"Semantic Search is not Keyword Search.","target":"keyword-search","targetLabel":"Keyword Search","targetUri":"https://id.searchplex.net/keyword-search/"},{"reason":"Semantic search is a retrieval capability or product term; RAG is a generation pattern that may use semantic search.","statement":"Semantic Search is not Retrieval-Augmented Generation.","target":"rag","targetLabel":"Retrieval-Augmented Generation","targetUri":"https://id.searchplex.net/rag/"},{"reason":"Semantic search describes meaning-oriented retrieval; vector search describes a mechanism based on vector similarity.","statement":"Semantic Search is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"semantic-search","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Semantic Search","type":"canonicalSurvey","url":"https://link.springer.com/referenceworkentry/10.1007/978-3-319-63962-8_154-1"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Semantic Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Semantic search may use dense retrieval, but it is a broader product or intent concept rather than a specific representation family.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Semantic search retrieves by meaning; grounding concerns whether output is supported by source material.","target":"grounding","type":"notEquivalentTo"},{"note":"Semantic search describes meaning-oriented retrieval; keyword search describes retrieval centered on explicit query keywords or phrases.","target":"keyword-search","type":"notEquivalentTo"},{"note":"Semantic search is a retrieval capability or product term; RAG is a generation pattern that may use semantic search.","target":"rag","type":"notEquivalentTo"},{"note":"Semantic search describes meaning-oriented retrieval; vector search describes a mechanism based on vector similarity.","target":"vector-search","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"grounding","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"image-retrieval","type":"related"},{"target":"keyword-search","type":"related"},{"target":"learned-sparse-retrieval","type":"related"},{"target":"multimodal-retrieval","type":"related"},{"target":"rag","type":"related"},{"target":"vector-search","type":"related"},{"target":"vocabulary-mismatch","type":"related"}],"representations":{"html":"https://id.searchplex.net/semantic-search/","json":"https://id.searchplex.net/semantic-search.json","jsonld":"https://id.searchplex.net/semantic-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Semantic search does not denote one retrieval algorithm or representation. Dense retrieval, learned sparse retrieval, entity-based methods, query expansion, and hybrid retrieval may all contribute to systems described as semantic search.","status":"published","terms":[{"community":"industry","label":"Semantic Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/semantic-search/"}
{"altLabels":["Search Engine Results Page","Search Results Page"],"definition":"A search engine results page displayed in response to a user query.","derivedRelations":[{"note":"Web search is the task or system setting; a SERP is the presented search-results page.","target":"web-search","type":"notEquivalentTo"},{"target":"diversified-retrieval","type":"related"},{"target":"session-search","type":"related"},{"target":"web-search","type":"related"}],"distinctions":[{"reason":"Ranking orders results; a SERP is the page or interface where ranked and other search results are presented.","statement":"SERP is not Ranking.","target":"ranking","targetLabel":"Ranking","targetUri":"https://id.searchplex.net/ranking/"},{"reason":"Retrieval selects items; a SERP presents retrieved results and related search features to the user.","statement":"SERP is not Retrieval.","target":"retrieval","targetLabel":"Retrieval","targetUri":"https://id.searchplex.net/retrieval/"}],"id":"serp","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Google Search documentation","type":"officialHomepage","url":"https://developers.google.com/search/docs/fundamentals/how-search-works"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"SERP","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Ranking orders results; a SERP is the page or interface where ranked and other search results are presented.","target":"ranking","type":"notEquivalentTo"},{"note":"Retrieval selects items; a SERP presents retrieved results and related search features to the user.","target":"retrieval","type":"notEquivalentTo"},{"target":"faceted-search","type":"related"},{"target":"interactive-search","type":"related"},{"target":"ranking","type":"related"},{"target":"retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/serp/","json":"https://id.searchplex.net/serp.json","jsonld":"https://id.searchplex.net/serp.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"SERP refers to the result presentation surface, not the retrieval or ranking task itself. A SERP may include organic results, sponsored results, snippets, facets, answer boxes, entity panels, or other search features.","status":"published","uri":"https://id.searchplex.net/serp/"}
{"altLabels":["Search Session"],"definition":"A search task or setting that considers a sequence of related user interactions within a search session rather than a single isolated query.","derivedRelations":[{"note":"Single-shot retrieval treats a query or topic in isolation; session search uses multiple related interactions within a search session.","target":"single-shot-retrieval","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"iterative-retrieval","type":"related"}],"distinctions":[{"reason":"Session search considers multiple interactions over time; ad hoc retrieval is often framed as a single query against a fixed collection.","statement":"Session Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Session search uses session context across interactions; conversational search specifically involves a conversational interface or dialogue.","statement":"Session Search is not Conversational Search.","target":"conversational-search","targetLabel":"Conversational Search","targetUri":"https://id.searchplex.net/conversational-search/"}],"id":"session-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Interactive Information Retrieval","type":"canonicalSurvey","url":"https://doi.org/10.1561/1500000023"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Session Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"interactive-search","type":"broader"},{"note":"Session search considers multiple interactions over time; ad hoc retrieval is often framed as a single query against a fixed collection.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Session search uses session context across interactions; conversational search specifically involves a conversational interface or dialogue.","target":"conversational-search","type":"notEquivalentTo"},{"target":"conversational-search","type":"related"},{"target":"query-autocomplete","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-suggestion","type":"related"},{"target":"serp","type":"related"},{"target":"single-shot-retrieval","type":"related"}],"representations":{"html":"https://id.searchplex.net/session-search/","json":"https://id.searchplex.net/session-search.json","jsonld":"https://id.searchplex.net/session-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Session search may use earlier queries, clicks, reformulations, dwell behavior, or other session context to support later retrieval. It is broader than conversational search because a session need not be dialogue-based.","status":"published","uri":"https://id.searchplex.net/session-search/"}
{"altLabels":["One-Shot Retrieval","Single Shot Retrieval","One Shot Retrieval"],"definition":"A non-iterative retrieval setup in which a single query, topic, or information need is submitted and a result set or ranked list is produced without dependent follow-up retrieval steps.","distinctions":[{"reason":"Ad hoc retrieval is a classic task formulation that is often evaluated in a single-shot way, but single-shot retrieval names the non-iterative interaction pattern rather than the collection/topic setup.","statement":"Single-Shot Retrieval is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Single-shot retrieval is non-iterative; agentic retrieval lets an agent plan, issue, evaluate, and iterate retrieval actions.","statement":"Single-Shot Retrieval is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"Single-shot retrieval produces results without user-system interaction over time; interactive search centers on interaction across a search process.","statement":"Single-Shot Retrieval is not Interactive Search.","target":"interactive-search","targetLabel":"Interactive Search","targetUri":"https://id.searchplex.net/interactive-search/"},{"reason":"Single-shot retrieval uses one non-iterative retrieval attempt; iterative retrieval repeats retrieval attempts.","statement":"Single-Shot Retrieval is not Iterative Retrieval.","target":"iterative-retrieval","targetLabel":"Iterative Retrieval","targetUri":"https://id.searchplex.net/iterative-retrieval/"},{"reason":"Single-shot retrieval uses one retrieval attempt or query; multi-hop retrieval uses multiple dependent retrieval steps or evidence hops.","statement":"Single-Shot Retrieval is not Multi-Hop Retrieval.","target":"multi-hop-retrieval","targetLabel":"Multi-Hop Retrieval","targetUri":"https://id.searchplex.net/multi-hop-retrieval/"},{"reason":"Single-shot retrieval treats a query or topic in isolation; session search uses multiple related interactions within a search session.","statement":"Single-Shot Retrieval is not Session Search.","target":"session-search","targetLabel":"Session Search","targetUri":"https://id.searchplex.net/session-search/"}],"id":"single-shot-retrieval","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"An example information retrieval problem","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Single-Shot Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Ad hoc retrieval is a classic task formulation that is often evaluated in a single-shot way, but single-shot retrieval names the non-iterative interaction pattern rather than the collection/topic setup.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Single-shot retrieval is non-iterative; agentic retrieval lets an agent plan, issue, evaluate, and iterate retrieval actions.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"Single-shot retrieval produces results without user-system interaction over time; interactive search centers on interaction across a search process.","target":"interactive-search","type":"notEquivalentTo"},{"note":"Single-shot retrieval uses one non-iterative retrieval attempt; iterative retrieval repeats retrieval attempts.","target":"iterative-retrieval","type":"notEquivalentTo"},{"note":"Single-shot retrieval uses one retrieval attempt or query; multi-hop retrieval uses multiple dependent retrieval steps or evidence hops.","target":"multi-hop-retrieval","type":"notEquivalentTo"},{"note":"Single-shot retrieval treats a query or topic in isolation; session search uses multiple related interactions within a search session.","target":"session-search","type":"notEquivalentTo"},{"target":"ad-hoc-retrieval","type":"related"},{"target":"agentic-retrieval","type":"related"},{"target":"interactive-search","type":"related"},{"target":"iterative-retrieval","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"session-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/single-shot-retrieval/","json":"https://id.searchplex.net/single-shot-retrieval.json","jsonld":"https://id.searchplex.net/single-shot-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Single-shot retrieval is useful as a contrast with multi-hop retrieval, interactive search, session search, and agentic retrieval. The phrase one-shot retrieval is sometimes used informally, but single-shot avoids confusion with one-shot learning in machine learning.","status":"published","terms":[{"community":"industry","label":"Single-Turn Retrieval","usage":"nearSynonym"},{"community":"academic","label":"Single-Query Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/single-shot-retrieval/"}
{"definition":"Retrieval in which each query, indexed item, passage, document, or chunk is represented by one vector for similarity comparison.","derivedRelations":[{"note":"Hybrid retrieval often includes dense single-vector retrieval, but hybrid retrieval names a combination of retrieval signals rather than the single-vector representation format itself.","target":"hybrid-retrieval","type":"notEquivalentTo"},{"target":"approximate-nearest-neighbor-search","type":"related"},{"target":"embedding","type":"related"},{"target":"hybrid-retrieval","type":"related"}],"distinctions":[{"reason":"Dense retrieval describes retrieval using dense representations; single-vector retrieval describes the representation cardinality used for each retrievable item.","statement":"Single-Vector Retrieval is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Dual encoders commonly produce single-vector representations, but the architecture and the representation format are distinct concepts.","statement":"Single-Vector Retrieval is not Dual Encoder.","target":"dual-encoder","targetLabel":"Dual Encoder","targetUri":"https://id.searchplex.net/dual-encoder/"},{"reason":"Single-vector retrieval represents each item with one vector; multi-vector retrieval represents a query, item, or both with multiple vectors.","statement":"Single-Vector Retrieval is not Multi-Vector Retrieval.","target":"multi-vector-retrieval","targetLabel":"Multi-Vector Retrieval","targetUri":"https://id.searchplex.net/multi-vector-retrieval/"},{"reason":"Semantic search names a meaning-oriented retrieval goal or product capability; single-vector retrieval names one possible representation format.","statement":"Single-Vector Retrieval is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Vector search names a vector-similarity mechanism; single-vector retrieval names a retrieval representation format.","statement":"Single-Vector Retrieval is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"single-vector-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction","type":"definingReference","url":"https://arxiv.org/abs/2112.01488"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Single-Vector Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Dense retrieval describes retrieval using dense representations; single-vector retrieval describes the representation cardinality used for each retrievable item.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Dual encoders commonly produce single-vector representations, but the architecture and the representation format are distinct concepts.","target":"dual-encoder","type":"notEquivalentTo"},{"note":"Single-vector retrieval represents each item with one vector; multi-vector retrieval represents a query, item, or both with multiple vectors.","target":"multi-vector-retrieval","type":"notEquivalentTo"},{"note":"Semantic search names a meaning-oriented retrieval goal or product capability; single-vector retrieval names one possible representation format.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Vector search names a vector-similarity mechanism; single-vector retrieval names a retrieval representation format.","target":"vector-search","type":"notEquivalentTo"},{"target":"dense-retrieval","type":"related"},{"target":"dual-encoder","type":"related"},{"target":"multi-vector-retrieval","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/single-vector-retrieval/","json":"https://id.searchplex.net/single-vector-retrieval.json","jsonld":"https://id.searchplex.net/single-vector-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Single-vector retrieval is often the default format in dense retrieval and vector-search systems, but it is not identical to dense retrieval, vector search, semantic search, or the dual-encoder architecture that commonly produces such vectors.","status":"published","terms":[{"community":"academic","label":"Single-Vector Dense Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/single-vector-retrieval/"}
{"definition":"Retrieval using high-dimensional representations in which most feature dimensions have zero weight.","derivedRelations":[{"target":"learned-sparse-retrieval","type":"narrower"},{"note":"Keyword search describes a search style using explicit terms or phrases; sparse retrieval describes a representation property and includes both traditional lexical and learned sparse approaches.","target":"keyword-search","type":"notEquivalentTo"},{"note":"Lexical retrieval is based on explicit lexical features; sparse retrieval is the broader representation pattern that also includes learned sparse models.","target":"lexical-retrieval","type":"notEquivalentTo"},{"target":"dynamic-pruning","type":"related"},{"target":"hybrid-retrieval","type":"related"},{"target":"keyword-search","type":"related"}],"distinctions":[{"reason":"BM25 is a traditional lexical scoring method with sparse lexical evidence; sparse retrieval is the broader representation family.","statement":"Sparse Retrieval is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"Sparse retrieval includes traditional lexical sparse representations as well as learned sparse representations; learned sparse retrieval is the neural learned subset.","statement":"Sparse Retrieval is not Learned Sparse Retrieval.","target":"learned-sparse-retrieval","targetLabel":"Learned Sparse Retrieval","targetUri":"https://id.searchplex.net/learned-sparse-retrieval/"}],"id":"sparse-retrieval","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"The term vocabulary and postings lists","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/the-term-vocabulary-and-postings-lists-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Sparse Retrieval","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BM25 is a traditional lexical scoring method with sparse lexical evidence; sparse retrieval is the broader representation family.","target":"bm25","type":"notEquivalentTo"},{"note":"Sparse retrieval includes traditional lexical sparse representations as well as learned sparse representations; learned sparse retrieval is the neural learned subset.","target":"learned-sparse-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"tf-idf","type":"related"},{"target":"vector-space-model","type":"related"}],"representations":{"html":"https://id.searchplex.net/sparse-retrieval/","json":"https://id.searchplex.net/sparse-retrieval.json","jsonld":"https://id.searchplex.net/sparse-retrieval.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Traditional lexical representations and learned sparse representations can both be sparse. Sparse retrieval therefore should not be used as a synonym for learned sparse retrieval.","status":"published","uri":"https://id.searchplex.net/sparse-retrieval/"}
{"altLabels":["Query Spelling Correction","Spell Correction"],"definition":"Detecting and correcting misspelled or mistyped query terms in order to improve query interpretation or retrieval.","derivedRelations":[{"target":"search-as-you-type","type":"related"}],"distinctions":[{"reason":"Spelling correction corrects a submitted or completed query; query autocomplete predicts completions from a partially typed prefix.","statement":"Spelling Correction is not Query Autocomplete.","target":"query-autocomplete","targetLabel":"Query Autocomplete","targetUri":"https://id.searchplex.net/query-autocomplete/"},{"reason":"Spelling correction changes likely misspelled terms; query relaxation loosens constraints or matching requirements.","statement":"Spelling Correction is not Query Relaxation.","target":"query-relaxation","targetLabel":"Query Relaxation","targetUri":"https://id.searchplex.net/query-relaxation/"},{"reason":"Spelling correction is a specific query-assistance task; query rewriting is the broader family of query transformations.","statement":"Spelling Correction is not Query Rewriting.","target":"query-rewriting","targetLabel":"Query Rewriting","targetUri":"https://id.searchplex.net/query-rewriting/"},{"reason":"Spelling correction proposes corrected forms of the user's query; query suggestion recommends related, alternative, or subsequent queries.","statement":"Spelling Correction is not Query Suggestion.","target":"query-suggestion","targetLabel":"Query Suggestion","targetUri":"https://id.searchplex.net/query-suggestion/"}],"id":"spelling-correction","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Spelling correction","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/spelling-correction-1.html"},{"label":"Query Spelling Correction","type":"definingReference","url":"https://link.springer.com/referenceworkentry/10.1007/978-3-030-58335-8_4"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Spelling Correction","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Spelling correction corrects a submitted or completed query; query autocomplete predicts completions from a partially typed prefix.","target":"query-autocomplete","type":"notEquivalentTo"},{"note":"Spelling correction changes likely misspelled terms; query relaxation loosens constraints or matching requirements.","target":"query-relaxation","type":"notEquivalentTo"},{"note":"Spelling correction is a specific query-assistance task; query rewriting is the broader family of query transformations.","target":"query-rewriting","type":"notEquivalentTo"},{"note":"Spelling correction proposes corrected forms of the user's query; query suggestion recommends related, alternative, or subsequent queries.","target":"query-suggestion","type":"notEquivalentTo"},{"target":"query-understanding","type":"partOf"},{"target":"query-autocomplete","type":"related"},{"target":"query-relaxation","type":"related"},{"target":"query-rewriting","type":"related"},{"target":"query-suggestion","type":"related"}],"representations":{"html":"https://id.searchplex.net/spelling-correction/","json":"https://id.searchplex.net/spelling-correction.json","jsonld":"https://id.searchplex.net/spelling-correction.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In search, spelling correction may automatically rewrite a query, offer a 'did you mean' suggestion, or decide to keep the original query. It is distinct from autocomplete and query suggestion because it responds to a completed or submitted query rather than predicting a prefix or recommending related searches.","status":"published","terms":[{"community":"industry","label":"Did You Mean","usage":"informal"},{"community":"commerce-search","label":"Typo Correction","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/spelling-correction/"}
{"altLabels":["Sparse Lexical and Expansion Model"],"artifactKind":"Model","definition":"A family of learned sparse neural retrieval models that produce sparse vocabulary-space representations for queries and documents.","derivedRelations":[{"note":"SPLADE is one model family that realizes learned sparse retrieval, not the whole representation family.","target":"learned-sparse-retrieval","type":"notEquivalentTo"}],"distinctions":[{"reason":"SPLADE is a learned sparse model family with expansion-like behavior; query expansion is the task of enriching a query representation.","statement":"SPLADE is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"}],"id":"splade","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"SPLADE paper","type":"definingReference","url":"https://doi.org/10.1145/3404835.3463098"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"SPLADE","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"SPLADE is a learned sparse model family with expansion-like behavior; query expansion is the task of enriching a query representation.","target":"query-expansion","type":"notEquivalentTo"},{"target":"learned-sparse-retrieval","type":"realizes"},{"target":"dynamic-pruning","type":"related"},{"note":"SPLADE has expansion-like behavior but is not the query-expansion task.","target":"query-expansion","type":"related"},{"target":"candidate-retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/splade/","json":"https://id.searchplex.net/splade.json","jsonld":"https://id.searchplex.net/splade.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"SPLADE learns sparse expansion-like activations and is commonly used for first-stage retrieval or ranking. It is an implementation/model family within learned sparse retrieval, not a synonym for either learned sparse retrieval generally or query expansion.","status":"published","terms":[{"community":"academic","label":"Sparse Expansion Model","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/splade/"}
{"altLabels":["Hit@K","Hits@K","Hit Rate@K"],"definition":"A cutoff evaluation metric indicating whether at least one relevant item appears within the top K results.","distinctions":[{"reason":"Success@K is binary per query; Precision@K measures the fraction of top-K results that are relevant.","statement":"Success@K is not Precision@K.","target":"precision-at-k","targetLabel":"Precision@K","targetUri":"https://id.searchplex.net/precision-at-k/"},{"reason":"Success@K asks whether at least one relevant item appears in the top K; Recall@K measures how much of the relevant set appears in the top K.","statement":"Success@K is not Recall@K.","target":"recall-at-k","targetLabel":"Recall@K","targetUri":"https://id.searchplex.net/recall-at-k/"}],"id":"success-at-k","kind":"Metric","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Hit Rate metric","type":"officialHomepage","url":"https://docs.futureagi.com/future-agi/products/evaluate/metrics/retrieval/hit-rate"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Success@K","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"cutoff-metric","type":"broader"},{"target":"recall-metric","type":"broader"},{"target":"ranking","type":"evaluates"},{"target":"retrieval","type":"evaluates"},{"note":"Success@K is binary per query; Precision@K measures the fraction of top-K results that are relevant.","target":"precision-at-k","type":"notEquivalentTo"},{"note":"Success@K asks whether at least one relevant item appears in the top K; Recall@K measures how much of the relevant set appears in the top K.","target":"recall-at-k","type":"notEquivalentTo"},{"target":"mrr","type":"related"},{"target":"precision-at-k","type":"related"},{"target":"qrels","type":"related"},{"target":"recall-at-k","type":"related"}],"representations":{"html":"https://id.searchplex.net/success-at-k/","json":"https://id.searchplex.net/success-at-k.json","jsonld":"https://id.searchplex.net/success-at-k.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Success@K is a binary per-query metric often averaged across queries. It is commonly called Hit@K or Hits@K in recommendation, RAG, and retrieval evaluation settings.","status":"published","uri":"https://id.searchplex.net/success-at-k/"}
{"altLabels":["IR Evaluation Dataset","Retrieval Evaluation Dataset","Search Evaluation Dataset","Evaluation Dataset"],"artifactKind":"Benchmark","definition":"An information-retrieval evaluation resource consisting of a document or item collection, a set of topics or queries, and relevance judgments.","derivedRelations":[{"target":"corpus","type":"hasPart"},{"target":"qrels","type":"hasPart"},{"note":"BrowseComp is a browsing-agent benchmark; a test collection is a fixed IR evaluation resource consisting of corpus, topics or queries, and relevance judgments.","target":"browsecomp","type":"notEquivalentTo"},{"note":"CLEF is an evaluation program and research venue; a test collection is an evaluation resource, some of which are created through CLEF labs.","target":"clef","type":"notEquivalentTo"},{"note":"FIRE is an evaluation program and meeting; a test collection is an evaluation resource produced or used by specific FIRE tasks.","target":"fire","type":"notEquivalentTo"},{"note":"INEX is an evaluation program; its test collections are evaluation resources used by INEX tasks.","target":"inex","type":"notEquivalentTo"},{"note":"ir_datasets is software and catalog infrastructure for working with test collections; a test collection is an evaluation resource.","target":"ir-datasets","type":"notEquivalentTo"},{"note":"MTEB is a benchmark suite aggregating many tasks and datasets; a test collection is an evaluation resource pattern.","target":"mteb","type":"notEquivalentTo"},{"note":"NTCIR is an evaluation program and conference series; a test collection is an evaluation resource produced or used by specific tasks.","target":"ntcir","type":"notEquivalentTo"},{"note":"TRECVID is an evaluation campaign; a test collection is an evaluation resource used by a specific track or task.","target":"trecvid","type":"notEquivalentTo"},{"target":"ad-hoc-retrieval","type":"related"},{"target":"arguana","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"clef","type":"related"},{"target":"fever","type":"related"},{"target":"fiqa-2018","type":"related"},{"target":"fire","type":"related"},{"target":"hotpotqa","type":"related"},{"target":"inex","type":"related"},{"target":"llm-as-judge","type":"related"},{"target":"mteb","type":"related"},{"target":"natural-questions","type":"related"},{"target":"nfcorpus","type":"related"},{"target":"ntcir","type":"related"},{"target":"scifact","type":"related"},{"target":"trec-covid","type":"related"},{"target":"trecvid","type":"related"}],"distinctions":[{"reason":"BEIR is a benchmark suite aggregating multiple retrieval datasets and test collections; a test collection is a general evaluation-resource pattern.","statement":"Test Collection is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"A test collection includes a corpus plus topics or queries and relevance judgments; a corpus is only the searchable collection component.","statement":"Test Collection is not Corpus.","target":"corpus","targetLabel":"Corpus","targetUri":"https://id.searchplex.net/corpus/"},{"reason":"Pooling is one method for constructing judgment sets used in test collections; it is not the full collection.","statement":"Test Collection is not Pooling.","target":"pooling","targetLabel":"Pooling","targetUri":"https://id.searchplex.net/pooling/"},{"reason":"Qrels are the relevance-judgment component of a test collection, not the entire evaluation collection.","statement":"Test Collection is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"TREC is an evaluation program and workshop series; a test collection is an evaluation resource.","statement":"Test Collection is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"test-collection","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Text REtrieval Conference","type":"officialHomepage","url":"https://trec.nist.gov/"},{"label":"ir_datasets catalog","type":"officialHomepage","url":"https://ir-datasets.com/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Test Collection","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BEIR is a benchmark suite aggregating multiple retrieval datasets and test collections; a test collection is a general evaluation-resource pattern.","target":"beir","type":"notEquivalentTo"},{"note":"A test collection includes a corpus plus topics or queries and relevance judgments; a corpus is only the searchable collection component.","target":"corpus","type":"notEquivalentTo"},{"note":"Pooling is one method for constructing judgment sets used in test collections; it is not the full collection.","target":"pooling","type":"notEquivalentTo"},{"note":"Qrels are the relevance-judgment component of a test collection, not the entire evaluation collection.","target":"qrels","type":"notEquivalentTo"},{"note":"TREC is an evaluation program and workshop series; a test collection is an evaluation resource.","target":"trec","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"ms-marco-passage-ranking","type":"related"},{"target":"pooling","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"retrieval","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/test-collection/","json":"https://id.searchplex.net/test-collection.json","jsonld":"https://id.searchplex.net/test-collection.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In Cranfield-style and TREC-style evaluation, a test collection provides the fixed materials needed for offline retrieval evaluation. It is broader than qrels alone and may be packaged as part of a benchmark suite.","status":"published","terms":[{"community":"industry","label":"Benchmark Dataset","usage":"nearSynonym"},{"community":"industry","label":"Eval Dataset","usage":"informal"}],"uri":"https://id.searchplex.net/test-collection/"}
{"altLabels":["Term Frequency-Inverse Document Frequency","Term Frequency–Inverse Document Frequency"],"definition":"A term-weighting method that combines term frequency with inverse document frequency to weight terms in documents, queries, or vector-space representations.","derivedRelations":[{"target":"bm25f","type":"related"}],"distinctions":[{"reason":"TF-IDF is a term-weighting method; BM25 is a probabilistic ranking function with term saturation and document-length normalization.","statement":"TF-IDF is not BM25.","target":"bm25","targetLabel":"BM25","targetUri":"https://id.searchplex.net/bm25/"},{"reason":"TF-IDF is one classical term-weighting method; lexical retrieval is the broader retrieval family based on lexical features.","statement":"TF-IDF is not Lexical Retrieval.","target":"lexical-retrieval","targetLabel":"Lexical Retrieval","targetUri":"https://id.searchplex.net/lexical-retrieval/"}],"id":"tf-idf","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Scoring, term weighting and the vector space model","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/scoring-term-weighting-and-the-vector-space-model-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"TF-IDF","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"TF-IDF is a term-weighting method; BM25 is a probabilistic ranking function with term saturation and document-length normalization.","target":"bm25","type":"notEquivalentTo"},{"note":"TF-IDF is one classical term-weighting method; lexical retrieval is the broader retrieval family based on lexical features.","target":"lexical-retrieval","type":"notEquivalentTo"},{"target":"bm25","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"sparse-retrieval","type":"related"},{"target":"vector-space-model","type":"related"},{"target":"ranking","type":"usedFor"},{"target":"retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/tf-idf/","json":"https://id.searchplex.net/tf-idf.json","jsonld":"https://id.searchplex.net/tf-idf.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"TF-IDF is commonly associated with vector space retrieval and classical lexical baselines. It is not the same as BM25, lexical retrieval generally, or sparse retrieval generally.","status":"published","uri":"https://id.searchplex.net/tf-idf/"}
{"altLabels":["Tip-of-the-Tongue Retrieval","Tip-of-the-Tongue Known-Item Retrieval","ToT Search","ToT Retrieval","ToT-KIR"],"definition":"A known-item search task in which the searcher is trying to re-find or identify a previously encountered item but cannot recall a precise name, title, identifier, or other direct lookup key.","distinctions":[{"reason":"Tip-of-the-tongue search seeks a specific previously encountered item; ad hoc retrieval usually seeks topically relevant material for a new information need.","statement":"Tip-of-the-Tongue Search is not Ad Hoc Retrieval.","target":"ad-hoc-retrieval","targetLabel":"Ad Hoc Retrieval","targetUri":"https://id.searchplex.net/ad-hoc-retrieval/"},{"reason":"Exact match depends on precise terms, values, or phrases; tip-of-the-tongue search is needed when the user lacks the exact identifier.","statement":"Tip-of-the-Tongue Search is not Exact Match.","target":"exact-match","targetLabel":"Exact Match","targetUri":"https://id.searchplex.net/exact-match/"},{"reason":"Tip-of-the-tongue search is a specialized form of known-item search where the searcher cannot reliably recall the item identifier.","statement":"Tip-of-the-Tongue Search is not Known-Item Search.","target":"known-item-search","targetLabel":"Known-Item Search","targetUri":"https://id.searchplex.net/known-item-search/"},{"reason":"Tip-of-the-tongue search identifies an item from partial clues; question answering returns an answer or answer-bearing response.","statement":"Tip-of-the-Tongue Search is not Question Answering.","target":"question-answering","targetLabel":"Question Answering","targetUri":"https://id.searchplex.net/question-answering/"}],"id":"tip-of-the-tongue-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC Tip-of-the-Tongue Track","type":"officialHomepage","url":"https://trec-tot.github.io/"},{"label":"Tip of the Tongue Known-Item Retrieval","type":"definingReference","url":"https://arxiv.org/abs/2101.07124"},{"label":"Overview of the TREC 2025 Tip-of-the-Tongue Track","type":"definingReference","url":"https://arxiv.org/abs/2601.20671"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Tip-of-the-Tongue Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"known-item-search","type":"broader"},{"note":"Tip-of-the-tongue search seeks a specific previously encountered item; ad hoc retrieval usually seeks topically relevant material for a new information need.","target":"ad-hoc-retrieval","type":"notEquivalentTo"},{"note":"Exact match depends on precise terms, values, or phrases; tip-of-the-tongue search is needed when the user lacks the exact identifier.","target":"exact-match","type":"notEquivalentTo"},{"note":"Tip-of-the-tongue search is a specialized form of known-item search where the searcher cannot reliably recall the item identifier.","target":"known-item-search","type":"notEquivalentTo"},{"note":"Tip-of-the-tongue search identifies an item from partial clues; question answering returns an answer or answer-bearing response.","target":"question-answering","type":"notEquivalentTo"},{"target":"interactive-search","type":"related"},{"target":"multimodal-retrieval","type":"related"},{"target":"query-reformulation","type":"related"},{"target":"query-understanding","type":"related"},{"target":"question-answering","type":"related"}],"representations":{"html":"https://id.searchplex.net/tip-of-the-tongue-search/","json":"https://id.searchplex.net/tip-of-the-tongue-search.json","jsonld":"https://id.searchplex.net/tip-of-the-tongue-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Tip-of-the-tongue search queries often contain partial memories, contextual clues, uncertain descriptions, comparisons, or remembered attributes rather than exact identifiers. It is a specialized form of known-item search and is distinct from exact match search and ordinary ad hoc topical retrieval.","status":"published","terms":[{"community":"industry","label":"Forgotten Item Search","usage":"informal"},{"community":"general","label":"Find the Thing I Can't Name","usage":"informal"}],"uri":"https://id.searchplex.net/tip-of-the-tongue-search/"}
{"altLabels":["Text REtrieval Conference"],"definition":"The Text REtrieval Conference, an evaluation program and workshop series supporting large-scale information-retrieval evaluation.","derivedRelations":[{"note":"CLEF and TREC are distinct evaluation programs and venues, though both support shared evaluation in information retrieval and information access.","target":"clef","type":"notEquivalentTo"},{"note":"FIRE and TREC are distinct evaluation programs, though FIRE was modeled as a regional counterpart to programs such as TREC, CLEF, and NTCIR.","target":"fire","type":"notEquivalentTo"},{"note":"NTCIR and TREC are distinct evaluation programs, though both organize shared information-retrieval evaluation tasks.","target":"ntcir","type":"notEquivalentTo"},{"note":"TREC-COVID is a specific test collection/challenge dataset; TREC is the broader evaluation program.","target":"trec-covid","type":"notEquivalentTo"},{"note":"TRECVID originated from the TREC video track but is a distinct video-retrieval evaluation campaign.","target":"trecvid","type":"notEquivalentTo"},{"target":"clef","type":"related"},{"target":"fire","type":"related"},{"target":"inex","type":"related"},{"target":"ir-datasets","type":"related"},{"target":"ntcir","type":"related"},{"target":"trec-covid","type":"related"},{"target":"trecvid","type":"related"}],"distinctions":[{"reason":"TREC is an evaluation program and workshop series; BEIR is a benchmark suite aggregating multiple retrieval datasets.","statement":"TREC is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"TREC may publish qrels for tracks; qrels are judgment files, not the TREC program itself.","statement":"TREC is not Qrels.","target":"qrels","targetLabel":"Qrels","targetUri":"https://id.searchplex.net/qrels/"},{"reason":"TREC is an evaluation program and workshop series; a test collection is an evaluation resource, some of which are created by TREC tracks.","statement":"TREC is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"}],"id":"trec","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Text REtrieval Conference","type":"officialHomepage","url":"https://trec.nist.gov/"},{"label":"NIST TREC overview","type":"officialHomepage","url":"https://www.nist.gov/itl/iad/mig/text-retrieval-conference-trec"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"TREC","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"TREC is an evaluation program and workshop series; BEIR is a benchmark suite aggregating multiple retrieval datasets.","target":"beir","type":"notEquivalentTo"},{"note":"TREC may publish qrels for tracks; qrels are judgment files, not the TREC program itself.","target":"qrels","type":"notEquivalentTo"},{"note":"TREC is an evaluation program and workshop series; a test collection is an evaluation resource, some of which are created by TREC tracks.","target":"test-collection","type":"notEquivalentTo"},{"target":"pooling","type":"related"},{"target":"qrels","type":"related"},{"target":"relevance-judgment","type":"related"},{"target":"test-collection","type":"related"}],"representations":{"html":"https://id.searchplex.net/trec/","json":"https://id.searchplex.net/trec.json","jsonld":"https://id.searchplex.net/trec.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"TREC organizes tracks, topics, collections, runs, and evaluation resources. It is not itself a metric, a qrels file, or a single test collection, though many test collections and qrels are produced through TREC tracks.","status":"published","uri":"https://id.searchplex.net/trec/"}
{"artifactKind":"Dataset","definition":"A COVID-19 biomedical literature retrieval test collection created through the TREC-COVID challenge.","distinctions":[{"reason":"TREC-COVID is a component evaluation dataset; BEIR is the benchmark suite that includes it.","statement":"TREC-COVID is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"TREC-COVID is a specific test collection/challenge dataset; TREC is the broader evaluation program.","statement":"TREC-COVID is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"trec-covid","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC-COVID","type":"officialHomepage","url":"https://ir.nist.gov/covidSubmit/"},{"label":"TREC-COVID in ir_datasets","type":"officialHomepage","url":"https://ir-datasets.com/beir.html#beir/trec-covid"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"TREC-COVID","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"TREC-COVID is a component evaluation dataset; BEIR is the benchmark suite that includes it.","target":"beir","type":"notEquivalentTo"},{"note":"TREC-COVID is a specific test collection/challenge dataset; TREC is the broader evaluation program.","target":"trec","type":"notEquivalentTo"},{"target":"beir","type":"related"},{"target":"document-retrieval","type":"related"},{"target":"qrels","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/trec-covid/","json":"https://id.searchplex.net/trec-covid.json","jsonld":"https://id.searchplex.net/trec-covid.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"TREC-COVID is a specific evaluation dataset/test collection and is also used as a component dataset in BEIR. It should not be confused with TREC as an evaluation program or BEIR as a benchmark suite.","status":"published","uri":"https://id.searchplex.net/trec-covid/"}
{"altLabels":["TREC Video Retrieval Evaluation","TREC Video Track"],"definition":"The TREC Video Retrieval Evaluation, an evaluation campaign for content-based video analysis and retrieval.","distinctions":[{"reason":"TRECVID is an evaluation campaign for video analysis and retrieval; multimodal retrieval is a broader retrieval task family.","statement":"TRECVID is not Multimodal Retrieval.","target":"multimodal-retrieval","targetLabel":"Multimodal Retrieval","targetUri":"https://id.searchplex.net/multimodal-retrieval/"},{"reason":"TRECVID is an evaluation campaign; a test collection is an evaluation resource used by a specific track or task.","statement":"TRECVID is not Test Collection.","target":"test-collection","targetLabel":"Test Collection","targetUri":"https://id.searchplex.net/test-collection/"},{"reason":"TRECVID originated from the TREC video track but is a distinct video-retrieval evaluation campaign.","statement":"TRECVID is not TREC.","target":"trec","targetLabel":"TREC","targetUri":"https://id.searchplex.net/trec/"}],"id":"trecvid","kind":"EvaluationProgram","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TRECVID","type":"officialHomepage","url":"https://trecvid.nist.gov/"},{"label":"NIST TRECVID overview","type":"officialHomepage","url":"https://www.nist.gov/itl/iad/mig/trec-video-retrieval-evaluation-trecvid"},{"label":"Overview of evaluated video retrieval tasks at TRECVID","type":"canonicalSurvey","url":"https://arxiv.org/abs/2306.13118"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"TRECVID","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"TRECVID is an evaluation campaign for video analysis and retrieval; multimodal retrieval is a broader retrieval task family.","target":"multimodal-retrieval","type":"notEquivalentTo"},{"note":"TRECVID is an evaluation campaign; a test collection is an evaluation resource used by a specific track or task.","target":"test-collection","type":"notEquivalentTo"},{"note":"TRECVID originated from the TREC video track but is a distinct video-retrieval evaluation campaign.","target":"trec","type":"notEquivalentTo"},{"target":"cross-modal-retrieval","type":"related"},{"target":"image-retrieval","type":"related"},{"target":"multimodal-retrieval","type":"related"},{"target":"test-collection","type":"related"},{"target":"trec","type":"related"}],"representations":{"html":"https://id.searchplex.net/trecvid/","json":"https://id.searchplex.net/trecvid.json","jsonld":"https://id.searchplex.net/trecvid.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"TRECVID began as a TREC video track and became an independent annual evaluation campaign. It has supported tasks such as video search, shot-boundary detection, concept or feature extraction, instance search, event detection, video summarization, and video-to-text description.","status":"published","uri":"https://id.searchplex.net/trecvid/"}
{"definition":"A broad practitioner term for retrieving items using vector representations and vector similarity or distance operations.","derivedRelations":[{"note":"Cross-modal retrieval may use vector search, but it is a task involving modality boundaries rather than a vector-search mechanism itself.","target":"cross-modal-retrieval","type":"notEquivalentTo"},{"note":"Vector search is a similarity-search mechanism over vectors; Dense Passage Retrieval is a passage-retrieval method that may use vector search.","target":"dense-passage-retrieval","type":"notEquivalentTo"},{"note":"Reverse image search may use vector search internally, but it is a user-facing image-query retrieval task.","target":"reverse-image-search","type":"notEquivalentTo"},{"note":"Vector search names a vector-similarity mechanism; single-vector retrieval names a retrieval representation format.","target":"single-vector-retrieval","type":"notEquivalentTo"},{"note":"Web search can use many retrieval and ranking methods; vector search is one class of vector-based retrieval mechanism.","target":"web-search","type":"notEquivalentTo"},{"target":"dense-passage-retrieval","type":"related"}],"distinctions":[{"reason":"Vector search may use approximate nearest neighbor methods, but vector search is the broader retrieval mechanism.","statement":"Vector Search is not Approximate Nearest Neighbor Search.","target":"approximate-nearest-neighbor-search","targetLabel":"Approximate Nearest Neighbor Search","targetUri":"https://id.searchplex.net/approximate-nearest-neighbor-search/"},{"reason":"Vector search is broader than dense retrieval and can include sparse, binary, multi-vector, or other vector representations.","statement":"Vector Search is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"Vector search is a retrieval mechanism over vector representations; an embedding is a vector representation.","statement":"Vector Search is not Embedding.","target":"embedding","targetLabel":"Embedding","targetUri":"https://id.searchplex.net/embedding/"},{"reason":"HNSW is one index/search method often used for vector search; it is not vector search itself.","statement":"Vector Search is not HNSW.","target":"hnsw","targetLabel":"HNSW","targetUri":"https://id.searchplex.net/hnsw/"},{"reason":"Vector search names a retrieval mechanism; semantic search names a user-facing retrieval goal or product capability.","statement":"Vector Search is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Vector search is a broad practitioner term, usually for embedding or nearest-neighbor retrieval; the Vector Space Model is a classical term-vector retrieval model.","statement":"Vector Search is not Vector Space Model.","target":"vector-space-model","targetLabel":"Vector Space Model","targetUri":"https://id.searchplex.net/vector-space-model/"}],"id":"vector-search","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Introduction to vector search","type":"officialHomepage","url":"https://www.elastic.co/search-labs/blog/introduction-to-vector-search"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Vector Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Vector search may use approximate nearest neighbor methods, but vector search is the broader retrieval mechanism.","target":"approximate-nearest-neighbor-search","type":"notEquivalentTo"},{"note":"Vector search is broader than dense retrieval and can include sparse, binary, multi-vector, or other vector representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"Vector search is a retrieval mechanism over vector representations; an embedding is a vector representation.","target":"embedding","type":"notEquivalentTo"},{"note":"HNSW is one index/search method often used for vector search; it is not vector search itself.","target":"hnsw","type":"notEquivalentTo"},{"note":"Vector search names a retrieval mechanism; semantic search names a user-facing retrieval goal or product capability.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Vector search is a broad practitioner term, usually for embedding or nearest-neighbor retrieval; the Vector Space Model is a classical term-vector retrieval model.","target":"vector-space-model","type":"notEquivalentTo"},{"target":"approximate-nearest-neighbor-search","type":"related"},{"target":"cross-modal-retrieval","type":"related"},{"target":"dense-retrieval","type":"related"},{"target":"embedding","type":"related"},{"target":"hnsw","type":"related"},{"target":"image-retrieval","type":"related"},{"target":"multi-vector-retrieval","type":"related"},{"target":"reverse-image-search","type":"related"},{"target":"semantic-search","type":"related"},{"target":"single-vector-retrieval","type":"related"},{"target":"vector-space-model","type":"related"}],"representations":{"html":"https://id.searchplex.net/vector-search/","json":"https://id.searchplex.net/vector-search.json","jsonld":"https://id.searchplex.net/vector-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"In current industry usage, vector search often refers specifically to dense or embedding-based nearest-neighbor retrieval, but the term itself can also encompass sparse, binary, multi-vector, or other vector representations. It should not be collapsed with the classical Vector Space Model merely because both use vector representations.","status":"published","terms":[{"community":"industry","label":"Vector Retrieval","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/vector-search/"}
{"altLabels":["VSM"],"definition":"A classical information-retrieval model that represents documents and queries as vectors in a term space and compares them using vector similarity.","distinctions":[{"reason":"The classical vector space model usually uses sparse term-weighted vectors; dense retrieval uses learned dense representations.","statement":"Vector Space Model is not Dense Retrieval.","target":"dense-retrieval","targetLabel":"Dense Retrieval","targetUri":"https://id.searchplex.net/dense-retrieval/"},{"reason":"The vector space model is a retrieval model over term-weighted vectors; an embedding is a learned vector representation that may be used in many retrieval or non-retrieval settings.","statement":"Vector Space Model is not Embedding.","target":"embedding","targetLabel":"Embedding","targetUri":"https://id.searchplex.net/embedding/"},{"reason":"The vector space model is a classical term-vector retrieval model; vector search is a broad practitioner term often used for retrieval over learned vector embeddings.","statement":"Vector Space Model is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"vector-space-model","kind":"Representation","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Scoring, term weighting and the vector space model","type":"canonicalSurvey","url":"https://nlp.stanford.edu/IR-book/html/htmledition/scoring-term-weighting-and-the-vector-space-model-1.html"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Vector Space Model","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"The classical vector space model usually uses sparse term-weighted vectors; dense retrieval uses learned dense representations.","target":"dense-retrieval","type":"notEquivalentTo"},{"note":"The vector space model is a retrieval model over term-weighted vectors; an embedding is a learned vector representation that may be used in many retrieval or non-retrieval settings.","target":"embedding","type":"notEquivalentTo"},{"note":"The vector space model is a classical term-vector retrieval model; vector search is a broad practitioner term often used for retrieval over learned vector embeddings.","target":"vector-search","type":"notEquivalentTo"},{"target":"lexical-retrieval","type":"related"},{"target":"sparse-retrieval","type":"related"},{"target":"tf-idf","type":"related"},{"target":"vector-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/vector-space-model/","json":"https://id.searchplex.net/vector-space-model.json","jsonld":"https://id.searchplex.net/vector-space-model.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"This ID uses Vector Space Model in the classical IR sense: sparse term-weighted query and document vectors, often using TF-IDF-style weights. It should not be read as a generic label for modern embedding search, vector databases, approximate nearest-neighbor search, or dense vector retrieval merely because those systems also operate over vectors.","status":"published","uri":"https://id.searchplex.net/vector-space-model/"}
{"altLabels":["Vocabulary Gap","Term Mismatch"],"definition":"A retrieval failure mode or challenge in which a query and relevant documents use different terms or expressions for the same or related concepts.","derivedRelations":[{"target":"document-expansion","type":"related"}],"distinctions":[{"reason":"Vocabulary mismatch is a gap between query and document wording; query drift is a harmful change in query meaning introduced by expansion or reformulation.","statement":"Vocabulary Mismatch is not Query Drift.","target":"query-drift","targetLabel":"Query Drift","targetUri":"https://id.searchplex.net/query-drift/"},{"reason":"Query expansion is one technique for addressing vocabulary mismatch; vocabulary mismatch is the underlying terminology gap.","statement":"Vocabulary Mismatch is not Query Expansion.","target":"query-expansion","targetLabel":"Query Expansion","targetUri":"https://id.searchplex.net/query-expansion/"}],"id":"vocabulary-mismatch","kind":"Phenomenon","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Remedies against the Vocabulary Gap in Information Retrieval","type":"canonicalSurvey","url":"https://arxiv.org/abs/1711.06004"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Vocabulary Mismatch","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"Vocabulary mismatch is a gap between query and document wording; query drift is a harmful change in query meaning introduced by expansion or reformulation.","target":"query-drift","type":"notEquivalentTo"},{"note":"Query expansion is one technique for addressing vocabulary mismatch; vocabulary mismatch is the underlying terminology gap.","target":"query-expansion","type":"notEquivalentTo"},{"target":"dense-retrieval","type":"related"},{"target":"learned-sparse-retrieval","type":"related"},{"target":"lexical-retrieval","type":"related"},{"target":"query-drift","type":"related"},{"target":"query-expansion","type":"related"},{"target":"semantic-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/vocabulary-mismatch/","json":"https://id.searchplex.net/vocabulary-mismatch.json","jsonld":"https://id.searchplex.net/vocabulary-mismatch.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Vocabulary mismatch is one motivation for query expansion, semantic search, learned sparse retrieval, dense retrieval, and controlled vocabularies. It is distinct from query drift, where expansion or transformation moves the query away from the intended information need.","status":"published","uri":"https://id.searchplex.net/vocabulary-mismatch/"}
{"definition":"A dynamic-pruning query-evaluation method that uses upper bounds on potential scores to avoid fully evaluating candidates that cannot enter the current top-k result set.","derivedRelations":[{"note":"Block-Max WAND is a WAND-family refinement using block-level upper bounds; WAND is the broader dynamic-pruning method.","target":"block-max-wand","type":"notEquivalentTo"}],"distinctions":[{"reason":"WAND is an algorithmic dynamic-pruning method; weakAnd is a product-specific Vespa query operator.","statement":"WAND is not weakAnd.","target":"weakand","targetLabel":"weakAnd","targetUri":"https://id.searchplex.net/weakand/"}],"id":"wand","kind":"Method","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"WAND paper","type":"definingReference","url":"https://doi.org/10.1145/1571941.1572114"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"WAND","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"dynamic-pruning","type":"broader"},{"note":"WAND is an algorithmic dynamic-pruning method; weakAnd is a product-specific Vespa query operator.","target":"weakand","type":"notEquivalentTo"},{"target":"block-max-wand","type":"related"},{"target":"weakand","type":"related"}],"representations":{"html":"https://id.searchplex.net/wand/","json":"https://id.searchplex.net/wand.json","jsonld":"https://id.searchplex.net/wand.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"WAND is an algorithmic concept. Product operators named after WAND may implement specific variants or expose different guarantees.","status":"published","uri":"https://id.searchplex.net/wand/"}
{"artifactKind":"Software","definition":"A Vespa query operator that performs accelerated OR-like retrieval using a WAND-style dynamic-pruning algorithm while integrating with text matching and linguistic processing.","distinctions":[{"reason":"weakAnd is a Vespa query operator; WAND is an algorithmic dynamic-pruning method.","statement":"weakAnd is not WAND.","target":"wand","targetLabel":"WAND","targetUri":"https://id.searchplex.net/wand/"}],"id":"weakand","kind":"Artifact","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"Vespa weakAnd documentation","type":"officialHomepage","url":"https://docs.vespa.ai/en/reference/query-language-reference.html#weakand"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"weakAnd","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"weakAnd is a Vespa query operator; WAND is an algorithmic dynamic-pruning method.","target":"wand","type":"notEquivalentTo"},{"target":"dynamic-pruning","type":"related"},{"target":"wand","type":"related"},{"target":"candidate-retrieval","type":"usedFor"}],"representations":{"html":"https://id.searchplex.net/weakand/","json":"https://id.searchplex.net/weakand.json","jsonld":"https://id.searchplex.net/weakand.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"weakAnd is product-specific terminology. Vespa also exposes a distinct wand operator with different input and retrieval characteristics. weakAnd should therefore not be used as a general synonym for WAND.","status":"published","terms":[{"community":"vendor","label":"Weak AND","usage":"vendorUsage"}],"uri":"https://id.searchplex.net/weakand/"}
{"altLabels":["Browsing Agent","Web Agent"],"definition":"An AI agent that uses a web browser or web-search tools to navigate pages, issue searches, inspect sources, and gather information for a task.","derivedRelations":[{"target":"browsecomp","type":"evaluatedBy"},{"target":"browsecomp","type":"related"}],"distinctions":[{"reason":"A web browsing agent is an agent role or system; agentic retrieval is the retrieval behavior it may perform.","statement":"Web Browsing Agent is not Agentic Retrieval.","target":"agentic-retrieval","targetLabel":"Agentic Retrieval","targetUri":"https://id.searchplex.net/agentic-retrieval/"},{"reason":"BrowseComp is a benchmark for evaluating browsing agents; a web browsing agent is the system being evaluated.","statement":"Web Browsing Agent is not BrowseComp.","target":"browsecomp","targetLabel":"BrowseComp","targetUri":"https://id.searchplex.net/browsecomp/"},{"reason":"Web search is a search task or system setting; a web browsing agent is an agent that can use web search and browsing actions.","statement":"Web Browsing Agent is not Web Search.","target":"web-search","targetLabel":"Web Search","targetUri":"https://id.searchplex.net/web-search/"}],"id":"web-browsing-agent","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"BrowseComp","type":"definingReference","url":"https://arxiv.org/abs/2504.12516"},{"label":"Introducing deep research","type":"officialHomepage","url":"https://openai.com/index/introducing-deep-research/"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Web Browsing Agent","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"A web browsing agent is an agent role or system; agentic retrieval is the retrieval behavior it may perform.","target":"agentic-retrieval","type":"notEquivalentTo"},{"note":"BrowseComp is a benchmark for evaluating browsing agents; a web browsing agent is the system being evaluated.","target":"browsecomp","type":"notEquivalentTo"},{"note":"Web search is a search task or system setting; a web browsing agent is an agent that can use web search and browsing actions.","target":"web-search","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"deep-research","type":"related"},{"target":"multi-hop-retrieval","type":"related"},{"target":"retrieval-planning","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"},{"target":"web-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/web-browsing-agent/","json":"https://id.searchplex.net/web-browsing-agent.json","jsonld":"https://id.searchplex.net/web-browsing-agent.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"A web browsing agent may support deep research, question answering, fact finding, form filling, or broader web automation. It is not the same as a search engine, web search task, or the evaluation benchmark used to measure browsing agents.","status":"published","terms":[{"community":"industry","label":"Browser Agent","usage":"nearSynonym"},{"community":"industry","label":"Web Research Agent","usage":"nearSynonym"}],"uri":"https://id.searchplex.net/web-browsing-agent/"}
{"altLabels":["Browsing Agent Evaluation","Web Agent Evaluation"],"definition":"Evaluation of agents that search, browse, inspect web sources, and synthesize answers or results from information found on the web.","derivedRelations":[{"target":"browsecomp","type":"evaluatedBy"},{"target":"browsecomp-plus","type":"evaluatedBy"}],"distinctions":[{"reason":"BEIR evaluates retrieval systems over fixed IR datasets; web browsing agent evaluation measures agents that search and browse the web or web-like corpora.","statement":"Web Browsing Agent Evaluation is not BEIR.","target":"beir","targetLabel":"BEIR","targetUri":"https://id.searchplex.net/beir/"},{"reason":"MTEB evaluates text embedding models across tasks; web browsing agent evaluation evaluates agent behavior in web information-seeking tasks.","statement":"Web Browsing Agent Evaluation is not MTEB.","target":"mteb","targetLabel":"MTEB","targetUri":"https://id.searchplex.net/mteb/"},{"reason":"Web browsing agent evaluation evaluates agent behavior across search and browsing actions; web search is a search task or system setting.","statement":"Web Browsing Agent Evaluation is not Web Search.","target":"web-search","targetLabel":"Web Search","targetUri":"https://id.searchplex.net/web-search/"}],"id":"web-browsing-agent-evaluation","kind":"GeneralConcept","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"BrowseComp","type":"definingReference","url":"https://arxiv.org/abs/2504.12516"},{"label":"BrowseComp-Plus","type":"definingReference","url":"https://arxiv.org/abs/2508.06600"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Web Browsing Agent Evaluation","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"note":"BEIR evaluates retrieval systems over fixed IR datasets; web browsing agent evaluation measures agents that search and browse the web or web-like corpora.","target":"beir","type":"notEquivalentTo"},{"note":"MTEB evaluates text embedding models across tasks; web browsing agent evaluation evaluates agent behavior in web information-seeking tasks.","target":"mteb","type":"notEquivalentTo"},{"note":"Web browsing agent evaluation evaluates agent behavior across search and browsing actions; web search is a search task or system setting.","target":"web-search","type":"notEquivalentTo"},{"target":"agentic-retrieval","type":"related"},{"target":"browsecomp","type":"related"},{"target":"browsecomp-plus","type":"related"},{"target":"deep-research","type":"related"},{"target":"web-browsing-agent","type":"related"},{"target":"web-search","type":"related"}],"representations":{"html":"https://id.searchplex.net/web-browsing-agent-evaluation/","json":"https://id.searchplex.net/web-browsing-agent-evaluation.json","jsonld":"https://id.searchplex.net/web-browsing-agent-evaluation.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Web browsing agent evaluation may measure source finding, persistence, answer accuracy, evidence use, citation fidelity, and multi-step browsing behavior. It is distinct from ordinary web search evaluation and from retrieval-only benchmark suites.","status":"published","uri":"https://id.searchplex.net/web-browsing-agent-evaluation/"}
{"altLabels":["Web Retrieval"],"definition":"A search task or system setting focused on retrieving and ranking information from web-scale or web-like collections for user queries.","derivedRelations":[{"note":"Web search is a search task or system setting; a web browsing agent is an agent that can use web search and browsing actions.","target":"web-browsing-agent","type":"notEquivalentTo"},{"note":"Web browsing agent evaluation evaluates agent behavior across search and browsing actions; web search is a search task or system setting.","target":"web-browsing-agent-evaluation","type":"notEquivalentTo"},{"target":"browsecomp","type":"related"},{"target":"web-browsing-agent","type":"related"},{"target":"web-browsing-agent-evaluation","type":"related"}],"distinctions":[{"reason":"Web search is a search setting and task family; semantic search is a broad approach or capability focused on meaning-based matching.","statement":"Web Search is not Semantic Search.","target":"semantic-search","targetLabel":"Semantic Search","targetUri":"https://id.searchplex.net/semantic-search/"},{"reason":"Web search is the task or system setting; a SERP is the presented search-results page.","statement":"Web Search is not SERP.","target":"serp","targetLabel":"SERP","targetUri":"https://id.searchplex.net/serp/"},{"reason":"Web search can use many retrieval and ranking methods; vector search is one class of vector-based retrieval mechanism.","statement":"Web Search is not Vector Search.","target":"vector-search","targetLabel":"Vector Search","targetUri":"https://id.searchplex.net/vector-search/"}],"id":"web-search","kind":"Task","license":"https://creativecommons.org/publicdomain/zero/1.0/","links":[{"label":"TREC 2013 Web Track Overview","type":"definingReference","url":"https://arxiv.org/abs/1401.6501"}],"modifiedAt":"2026-09-14T06:43:46Z","prefLabel":"Web Search","publisher":{"name":"Searchplex","url":"https://searchplex.net/"},"relations":[{"target":"retrieval","type":"broader"},{"note":"Web search is a search setting and task family; semantic search is a broad approach or capability focused on meaning-based matching.","target":"semantic-search","type":"notEquivalentTo"},{"note":"Web search is the task or system setting; a SERP is the presented search-results page.","target":"serp","type":"notEquivalentTo"},{"note":"Web search can use many retrieval and ranking methods; vector search is one class of vector-based retrieval mechanism.","target":"vector-search","type":"notEquivalentTo"},{"target":"document-retrieval","type":"related"},{"target":"known-item-search","type":"related"},{"target":"ranking","type":"related"},{"target":"serp","type":"related"}],"representations":{"html":"https://id.searchplex.net/web-search/","json":"https://id.searchplex.net/web-search.json","jsonld":"https://id.searchplex.net/web-search.jsonld"},"scheme":"https://id.searchplex.net/scheme/","scopeNote":"Web search includes information needs such as informational, navigational, and transactional queries and is commonly studied through large-scale web collections and tracks. It is not equivalent to semantic search, vector search, or a particular retrieval method.","status":"published","uri":"https://id.searchplex.net/web-search/"}