Question Answering
Part of the Query Understanding series by Daniel Tunkelang.
Overview
Many search queries are questions, and for these queries users often want a direct answer rather than a list of documents to sift through. Question answering systems detect when a query has a question structure or intent, identify the type of answer required — a single fact, a list, a procedure, a comparison — and then extract or generate that answer from the available content. The quality of the answer depends on both retrieval, finding the right source material, and the step that produces an answer from it. Generative approaches can synthesize answers from multiple sources but introduce the risk of producing plausible-sounding but incorrect content, making attribution and source citation important considerations alongside accuracy.
Note: Article content behind Medium paywall — accessible at source URL with Medium account.