PostgreSQL Introduces BM25 Search Algorithm, Potentially Replacing Elasticsearch for Most Applications

PostgreSQL's new BM25 search algorithm extension threatens to eliminate Elasticsearch for most applications by fixing keyword stuffing and relevance ranking issues while enabling hybrid search that combines exact keyword matching with AI-powered semantic understanding in a single database.

• PostgreSQL now supports BM25 search algorithm through the pg_textsearch extension, eliminating the need for separate search systems like Elasticsearch for most applications
• BM25 fixes major problems with native PostgreSQL search including keyword stuffing, document length bias, and poor relevance ranking by using term frequency saturation and inverse document frequency
• The extension enables hybrid search combining BM25 with vector search for AI agents and RAG pipelines, providing both exact keyword matching and semantic understanding in a single database