Reciprocal rank fusion
A method for combining multiple result lists into one ranking by relative rank, without calibrating score scales between the methods.
On this page
Source content checked . English translation published 2026-10-27.
In business automation
Reciprocal rank fusion combines result lists such as lexical search and vector k-nearest-neighbor search by relative rank, without calibrating their score scales.
Examples
These examples illustrate a possible workflow. Adapt the connections and rules to your own systems.
Hybrid retrieval
Combine lexical search and vector retrieval into one ranked list before passing context to a model.
Relevant tools
- LlamaIndex
- Use reciprocal rank fusion when a retrieval pipeline needs to combine more than one ranked result list.
What to check
- The k parameter changes lower-rank weights, and the method needs at least two retrievers.