24th of 46 Questions.
What is MMR (Maximal Marginal Relevance) retrieval and how does it balance relevance with diversity of results?
Maximal Marginal Relevance (MMR) balances query relevance and diversity among retrieved documents, reducing redundancy in search results.
Standard similarity search often returns very similar or duplicate documents. MMR addresses this by selecting documents that are both relevant to the query and dissimilar from each other. It uses a lambda_mult parameter (0 to 1) to trade off relevance (closer to 1) versus diversity (closer to 0).
MMR Retrieval with LangChain
Key MMR Concepts
fetch_k: Number of documents initially retrieved; larger pools improve MMR’s selection ability.lambda_mult: Controls relevance-diversity trade-off (0 = max diversity, 1 = max relevance).When to use MMR: For summarization, QA over redundant datasets, or generating varied options.