Filtered Vector Search: What Acorn Fixes, and What Fixes Acorn
Qdrant introduces new methods to solve filtered vector search issues. ACORN and filterable HNSW enhance performance.
Filtered vector search encounters issues when metadata filters convert a healthy nearest-neighbor graph into scattered islands. Qdrant addresses this through two methods: Filterable HNSW adds extra edges during indexing, while ACORN steps through neighbors of neighbors during search. These approaches allow the query planner to select the most efficient strategy.