/

Apache Solr Improves Neural Search Capability

Apache Solr Improves Neural Search Capability

Apache Solr 9.1 Adds KNN, HSNW Filtering, and other Neural Search Capabilities

In the never-ending race to enable “intelligent” search, Apache Solr had been running behind Elasticsearch and OpenSearch – until now. The latest release of Apache Solr 9.1 includes improvements in KNN (K-Nearest Neighbor), and HSNW filtering (Hierarchical Navigable Small World) – features generally described under the broader umbrella of “Neural Search”.

From our initial experience, this is a huge improvement.  Joined filter queries with vector searches are now performing well. Prior to SOLR 9.1 it was a post filter (very inefficient), now finally it’s a pre-filter (in-filter).

This is a video from a respected colleague – Alessandro Benedetti – who implemented the work. He also speaks about some upcoming changes (for vector search) in his later slides.  Big kudos to Alessandro for his impressive continuing contributions, to the benefit of the whole Solr community.

We think neural / cognitive search is key to implementing hybrid search applications which combine the best of established keyword search methodologies and newer, AI-driven search technologies.

Whether you are implementing a new search application or migrating to open source, Apache Solr remains a viable solution.  Pureinsights offers a wide range of consulting services, complementary technologies, and managed services to help you with your implementation.

If you would have any questions about the new developments in Solr, or would like to set up a free initial consultation, please CONTACT US.

Related resources:

Stay up to date with our latest insights!