Elasticsearch-Quality Full Text Search in Postgres
Presented by:

Philippe Noël
Co-founder, CEO @ ParadeDB. ParadeDB is an Elasticsearch alternative built on Postgres. We're modernizing the features of Elasticsearch's product suite, starting with real-time search and analytics.
This talk will cover the current state of full-text search in Postgres (LIKE, ts_vector + GIN, pg_trgm) and motivations for what is missing and why it needs to be improved.
It will then discuss pg_search, an open-source Postgres extension that improves full-text search in Postgres by integrating Tantivy, a Lucene-inspired search library, to offer a new type of index. We'll cover the BM25 index, our integration with Postgres at the query planner via Postgres custom scans, and how we improve upon columnar aggregates over full-text search results (i.e. facets).
The talk will conclude with future areas of work in Postgres search and analytics.
- Date:
- 2025 March 19 14:00 EDT
- Duration:
- 50 min
- Room:
- Gold Coast 1&2
- Conference:
- Postgres Conference 2025
- Language:
- Track:
- Dev
- Difficulty:
- Medium