Searchplex ID Retrieval terminology registry

Task

Question Answering

Canonical ID https://id.searchplex.net/question-answering/

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.

Also called

  • QA
  • Question-Answering

Scope

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.

Observed terms

  • Answer Search industry
  • Answer Generation ml-nlp

Part of family

Not equivalent to

  • Question Answering is not Document Retrieval. Question answering aims to return an answer or answer-bearing response; document retrieval returns documents for a query or information need.
  • Question Answering is not Grounding. Question answering is a task objective; grounding concerns whether an answer or generated response is supported by source material.
  • Question Answering is not Retrieval-Augmented Generation. Question answering is a task; RAG is a retrieval-augmented generation pattern that can be used for question answering and other tasks.
  • Question Answering is not Deep Research. Question answering returns an answer to a question; deep research often produces a synthesized, cited research result after multiple search and reasoning steps.
  • Question Answering is not FiQA-2018. FiQA-2018 is a dataset used for financial QA/retrieval evaluation; question answering is the task.
  • Question Answering is not HotpotQA. HotpotQA is a dataset; question answering is the task.
  • Question Answering is not Natural Questions. Natural Questions is a dataset; question answering is the task.
  • Question Answering is not Tip-of-the-Tongue Search. Tip-of-the-tongue search identifies an item from partial clues; question answering returns an answer or answer-bearing response.

Learn more

ID