Google is a leader in The Forrester Wave™: Public Cloud Platforms, Q3 2026
cloud.google.com·4d ago
Cloud & Infrastructure
Google Cloud Blog·16h ago·5 min read
Announcing Native BM25 Ranking in AlloyDB and Cloud SQL
TL;DR
Google Cloud has introduced native BM25 ranking in AlloyDB and Cloud SQL, enhancing search capabilities by combining semantic vector search with traditional keyword search. This hybrid approach improves the handling of specific alphanumeric IDs and product SKUs.
✦ Why It Matters
Start using native BM25 ranking in your search applications to improve accuracy with alphanumeric queries.
Key Takeaways
- Enable pg_textsearch for BM25 ranking in your database.
- Create both vector and keyword indexes on the same table.
- Utilize the hybrid search UDF for unified search results.
- Experience faster vector search queries with AlloyDB.
How It Works
The native BM25 ranking integrates with PostgreSQL through the pg_textsearch extension, allowing users to create a unified search backend. This backend supports both semantic vector search and traditional keyword search, enabling more accurate and relevant results.
The hybrid search user-defined function (UDF) merges results from both search types using the Reciprocal Rank Fusion (RRF) algorithm, ensuring that users receive the best matches based on their queries.