RRF
A multi-source retrieval result fusion algorithm that computes a weighted sum of reciprocal ranks from different retrieval systems without requiring score normalization. Its simplicity and robustness make it a common choice for merging results in hybrid retrieval pipelines.
Core Facts
混合检索（Hybrid Search）将向量检索与BM25关键词检索的结果通过RRF（倒数排名融合）等算法合并，兼顾语义理解与精确匹配
Hybrid retrieval typically merges results from two retrieval pipelines using a weighted combination via the RRF (Reciprocal Rank Fusion) algorithm.
RRF does not require normalization of raw scores from different retrievers and relies only on rank information, but when multiple retrievers systematically make the same type of error, RRF only amplifies the mistakes
Timeline (last 90 days)
RRF uses only ranks rather than raw scores, avoiding the problem that the dimensions of dense vectors and sparse BM25 are not comparable.
The RRF fusion formula is score += weight / (k + rank + 1), with k_constant set to 60, dense weight 0.7, and sparse weight 0.3.
The retrieval pipeline uses a four-stage pipeline: parallel dense + sparse retrieval, RRF fusion, cross-encoder reranking, and final filtering.
混合检索的四阶段流程为：第一阶段并行稠密检索和稀疏检索；第二阶段用RRF融合，k_constant默认60，稠密权重0.7，稀疏权重0.3；第三阶段用Cross-Encoder重排序（推荐ms-marco-MiniLM-L-6-v2）；第四阶段截取Top-K，默认k=5，fetch_k=20。
RRF（倒数排名融合）是一种无需训练、无需归一化的多路检索结果融合算法，公式为score = Σ weight_r / (k + rank_r(d))，k通常取60，被Elasticsearch和Weaviate等系统采纳为默认融合策略。
For most enterprise document scenarios, static RRF is already sufficiently robust and is the lowest-cost starting point for adopting hybrid retrieval.
Reciprocal Rank Fusion (RRF) computes the final score for each document by summing the reciprocal of its rank in each ranked list plus a smoothing constant k (typically set to 60).
The core formula of RRF is RRF_score(d) = Σ 1 / (k + rank_i(d)), where k is a smoothing constant commonly set to 60.
RRF (Reciprocal Rank Fusion) weights and sums the results of two retrieval streams by the reciprocal of their ranks, requiring no score normalization, and is simple to implement with strong robustness.
When the number of knowledge bases is small, fusion methods such as RRF work well, but when the number of knowledge bases grows beyond 10, the default fusion strategy breaks down.
2 more timeline events
All Facts (17)
混合检索（Hybrid Search）将向量检索与BM25关键词检索的结果通过RRF（倒数排名融合）等算法合并，兼顾语义理解与精确匹配
90%VerifiedHybrid retrieval typically merges results from two retrieval pipelines using a weighted combination via the RRF (Reciprocal Rank Fusion) algorithm.
80%VerifiedRRF does not require normalization of raw scores from different retrievers and relies only on rank information, but when multiple retrievers systematically make the same type of error, RRF only amplifies the mistakes
65%VerifiedHybrid retrieval, which fuses BM25 keyword scores with vector similarity scores through weighted combination (typically using the RRF reciprocal rank fusion algorithm), has become the mainstream paradigm for RAG systems.
65%VerifiedHybrid retrieval combines BM25 sparse retrieval with dense vector retrieval, merging ranking results via RRF or linear weighting.
65%VerifiedThe Reciprocal Rank Fusion (RRF) algorithm was proposed by Cormack et al. in 2009, with a core formula of Score = Σ 1/(k + rank_i), where k typically takes the value 60 as a smoothing constant.
65%UnverifiedRRF对各路检索的原始分数分布不敏感，相比简单的线性加权在混合检索场景中更加鲁棒
85%UnverifiedRRF uses only ranks rather than raw scores, avoiding the problem that the dimensions of dense vectors and sparse BM25 are not comparable.
50%UnverifiedThe retrieval pipeline uses a four-stage pipeline: parallel dense + sparse retrieval, RRF fusion, cross-encoder reranking, and final filtering.
50%UnverifiedThe RRF fusion formula is score += weight / (k + rank + 1), with k_constant set to 60, dense weight 0.7, and sparse weight 0.3.
50%UnverifiedRRF（倒数排名融合）是一种无需训练、无需归一化的多路检索结果融合算法，公式为score = Σ weight_r / (k + rank_r(d))，k通常取60，被Elasticsearch和Weaviate等系统采纳为默认融合策略。
50%Unverified混合检索的四阶段流程为：第一阶段并行稠密检索和稀疏检索；第二阶段用RRF融合，k_constant默认60，稠密权重0.7，稀疏权重0.3；第三阶段用Cross-Encoder重排序（推荐ms-marco-MiniLM-L-6-v2）；第四阶段截取Top-K，默认k=5，fetch_k=20。
50%UnverifiedFor most enterprise document scenarios, static RRF is already sufficiently robust and is the lowest-cost starting point for adopting hybrid retrieval.
50%UnverifiedReciprocal Rank Fusion (RRF) computes the final score for each document by summing the reciprocal of its rank in each ranked list plus a smoothing constant k (typically set to 60).
50%UnverifiedThe core formula of RRF is RRF_score(d) = Σ 1 / (k + rank_i(d)), where k is a smoothing constant commonly set to 60.
50%UnverifiedRRF (Reciprocal Rank Fusion) weights and sums the results of two retrieval streams by the reciprocal of their ranks, requiring no score normalization, and is simple to implement with strong robustness.
50%UnverifiedWhen the number of knowledge bases is small, fusion methods such as RRF work well, but when the number of knowledge bases grows beyond 10, the default fusion strategy breaks down.
50%