# Hybrid Retrieval Hybrid retrieval combines more than one retrieval method—commonly lexical search, metadata filters, graph traversal, symbolic indexes, and vector similarity—then merges or reranks the candidates. The symbolic language engine already used several specialized retrieval geometries. A modern extension can add embeddings without discarding pronunciation indexes, taxonomic relations, trigger weights, source attestation, or explicit provenance. Related: [[wiki/Semantic Retrieval|Semantic Retrieval]], [[wiki/Vector Retrieval|Vector Retrieval]], [[wiki/Candidate Generation and Ranking|Candidate Generation and Ranking]], [[wiki/Lexical Knowledge Graph|Lexical Knowledge Graph]].