RRF ranker

Reciprocal Rank Fusion (RRF) is a simple method to combine multiple search results with different search score numerical distributions into a single list.

The main benefit of RRF is that it's lightweight and requires no tuning, but might provide less relevant results compared to other more computationally intensive ranking methods like Learn-to-Rank and Cross-Encoders.

The rrf rank operator takes two or more child sub-queries:

{
  "query": {
    "rrf": {
      "queries": [
        {"match": {"title": "cookie"}},
        {"semantic": {"title": "cookie"}}
      ],
      "k": 60.0,
      "rank_window_size": 20
    }
  }
}

And combines their score in the following way (taken from Elasticsearch docs):

score = 0.0
for q in queries:
    if d in result(q):
        score += 1.0 / ( k + rank( result(q), d ) )
return score

# where
# k is a ranking constant
# q is a query in the set of queries
# d is a document in the result set of q
# result(q) is the result set of q
# rank( result(q), d ) is d's rank within the result(q) starting from 1

Fields:

• queries (required, list of search queries). Two or more nested search queries to combine.
• k (optional, float, default 60.0). The ranking constant - how strongly lower document position affects the score.
• rank_window_size (optional, integer, default is request.size) This value determines the size of the individual result sets per query. A higher value will improve result relevance at the cost of performance. The final ranked result set is pruned down to the search request’s size.

The RRF ranker supports filters and aggregations, but does not support sorting.