Public demo

Dense Retrieval

A source-grounded theme page on Dense Passage Retrieval: dual encoders, semantic search, open-domain QA, and retrieval limits.

Theme overview

This theme covers the retrieval layer beneath many question answering and RAG systems. dense-passage-retrieval shows how learned vector representations can retrieve passages by semantic similarity rather than only exact lexical overlap.

Takeaways

• Dense retrieval is useful for finding related evidence when the user's wording differs from the paper.
• It does not remove the need for exact source inspection, especially in academic writing.
• SAV's search should combine semantic retrieval with citation-aware verification.

Included papers

• Dense Passage Retrieval for Open-Domain Question Answering

Evidence examples

• Dense Passage Retrieval for Open-Domain Question Answering: Dense passage retrieval learns embedding functions to map question-passage pairs into a vector space, enabling efficient retrieval of relevant passages for open-domain question answering.

  In this paper, we address the question: can we train a better dense embedding model using only pairs of questions and passages (or answers), without additional pretraining?

• Dense Passage Retrieval for Open-Domain Question Answering: Effective open-domain QA relies on passage retrieval to narrow search space, with dense encodings offering flexible, learnable representations superior to sparse methods.

  Open-domain question answering (QA) (Voorhees, 1999) is a task that answers factoid questions using a large collection of documents.