Hybrid Search
A retrieval approach that combines keyword search and vector search, merging the results of both to compensate for the respective shortcomings of literal matching and semantic matching. It is generally more robust than either approach alone.
Core Facts
Timeline (last 90 days)
At the end, the course points out the known limitations of the current hybrid search approach: spelling tolerance is not included in the scheme; expanding semantic coverage can only be done by increasing the number of embedded articles; long articles need to be split into small chunks and then embedded to reach content beyond the beginning; a re-ranker can be introduced to re-rank the top results.
Hybrid search runs dense vector search and classic keyword search (BM25) simultaneously, and typically uses Reciprocal Rank Fusion to merge the results, with the fusion formula summing 1 / (k + rank) over each document's rank in each list.
Hybrid search plus reranking is the upgrade every team pursuing better RAG should prioritize, rather than chasing fancier agent architectures.
When the evaluation set shows that vocabulary-mismatch errors dominate, hybrid search (keyword plus vector) has sufficient justification, and the benefit can be quantified.
All Facts (6)
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%Verified关键词搜索基于精确字符匹配无法理解同义词，语义搜索依赖向量嵌入可捕捉语义关联，两者结合的混合检索效果优于单一方法
65%UnverifiedAt the end, the course points out the known limitations of the current hybrid search approach: spelling tolerance is not included in the scheme; expanding semantic coverage can only be done by increasing the number of embedded articles; long articles need to be split into small chunks and then embedded to reach content beyond the beginning; a re-ranker can be introduced to re-rank the top results.
50%UnverifiedHybrid search runs dense vector search and classic keyword search (BM25) simultaneously, and typically uses Reciprocal Rank Fusion to merge the results, with the fusion formula summing 1 / (k + rank) over each document's rank in each list.
50%UnverifiedHybrid search plus reranking is the upgrade every team pursuing better RAG should prioritize, rather than chasing fancier agent architectures.
50%UnverifiedWhen the evaluation set shows that vocabulary-mismatch errors dominate, hybrid search (keyword plus vector) has sufficient justification, and the benefit can be quantified.
50%