Use HNSW index on Azure Cosmos DB for PostgreSQL for similarity search

In the previous post, you explored the IVFFlat (Inverted File with Flat Compression) index for approximate nearest neighbor search on Azure Cosmos DB for PostgreSQL. You observed that the IVFFlat index provides accurate results with lower search times compared to exact nearest neighbor search.

The pgvector extension provides another indexing algorithm for approximate nearest neighbor search called Hierarchical Navigable Small World (HNSW) graphs. HNSW is one of the most popular and best-performing indexes for vector similarity search. HNSW index …

ai approximate nearest neighbor azure azure cosmos db compression cosmos extension file hnsw index pgvector postgres postgresql results search