GlossaryRetrieval
Hybrid search
Combining keyword-based (BM25) and vector-based (embedding) retrieval into one ranked list.
Hybrid search runs both keyword (sparse, BM25) and semantic (dense, vector) retrieval over the same corpus and fuses the results via reciprocal rank fusion or learned weighting. Pure keyword search misses synonyms and paraphrase; pure semantic search misses exact-match scenarios (model names, codes, product SKUs). Hybrid covers both failure modes and is the modern default for serious RAG.