What Is Reciprocal Rank Fusion (RRF)?
A way to merge ranked result lists into one by scoring each item on its position in each list, not its raw score, so different search methods can be combined.
Hybrid search runs vector search and keyword search and gets two ranked lists back. Their scores can't be compared: a cosine similarity and a full-text score sit on different scales, so adding them lets one method drown out the other. Reciprocal rank fusion ignores the scores and uses only each item's position.
Each item earns 1 / (k + rank) from every list it appears in, and the totals set the final order. The constant k (60 in the original 2009 RRF paper by Cormack, Clarke and Büttcher) keeps first place ahead without making tenth place worthless. An item that ranks well in both lists beats one that tops only one of them.
RRF needs no training or score tuning, which is why several search engines offer it built in. It only merges, though: judging whether anything answers the question is the job of a reranker. For a longer walk-through, read Hybrid search explained.
How Kelu uses this
Kelu merges its pgvector and PostgreSQL full-text lists with weighted RRF, using k = 60. By default the vector list counts more than the full-text one, and lists from rewritten versions of the question count less than the user's own words, so a rewrite informs the ranking without outvoting what was asked. The fused list then goes to the reranker.
Related terms
Hybrid Search
A retrieval strategy that combines semantic (vector) search with keyword (full-text) search and merges the results, capturing both meaning and exact matches.
Reranking
A second pass in which a stronger model re-scores the top search results against the question, so the passages that actually answer it rise to the top.
Semantic Search
Search that ranks results by meaning similarity between the query and documents, using embeddings, rather than by matching keywords.
Embeddings
Numeric vector representations of text in which semantically similar passages end up close together, enabling search by meaning rather than exact words.
From definition to deployment
Connect a knowledge source and see cited AI answers on your own content in minutes — free, no credit card required.
Free to start · No credit card required