Hybrid Search

Hybrid search is a retrieval approach that combines keyword (lexical) search with semantic vector search, blending their results to improve relevance.

Keyword search excels at exact terms and rare tokens while semantic search captures meaning; combining them covers cases where either alone would miss results.

Hybrid search is widely used to feed more relevant context into RAG systems, often followed by a reranking step.