Weaviate 1.17: Hybrid Search with BM25 and Dense Vectors

AI Impact Summary

Weaviate 1.17 introduced hybrid search, combining BM25 and dense vector search for improved relevance. This approach leverages the strengths of both keyword-based and semantic search, offering a more robust solution for queries like "How to catch an Alaskan Pollock." The fusion algorithm, Reciprocal Rank Fusion (RRF), intelligently combines the ranked results from both search methods, ensuring a balanced representation of relevant documents and penalizing lower-ranked results.