Searchplex ID Retrieval terminology registry

Method

BM25

Canonical ID https://id.searchplex.net/bm25/

A term-based ranking function that scores documents for a query using term frequency, inverse document frequency, and document-length normalization.

Also called

  • Okapi BM25

Scope

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.

Used for

Not equivalent to

  • BM25 is not BM25F. BM25 is the standard BM25 scoring function; BM25F is a field-aware variant for structured or fielded documents.
  • BM25 is not Inverted Index. BM25 is a ranking function; an inverted index is an index structure commonly used to retrieve postings for lexical scoring.
  • BM25 is not Keyword Search. BM25 is a scoring method; keyword search is a practitioner-facing search style that may use BM25 or other lexical methods.
  • BM25 is not Lexical Retrieval. BM25 is one lexical scoring method; lexical retrieval is the broader retrieval family based on lexical features.
  • BM25 is not Sparse Retrieval. BM25 uses sparse lexical evidence, but sparse retrieval also includes other traditional and learned sparse approaches.
  • BM25 is not TF-IDF. BM25 and TF-IDF are both classical lexical weighting approaches, but BM25 adds probabilistic term saturation and document-length normalization.
  • BM25 is not Exact Match. BM25 scores documents using term statistics; exact match requires specified terms, values, or phrases to match exactly.

Learn more

ID