paradedb-skill

Implement Elasticsearch-quality full-text search in Postgres using the ParadeDB pg_search extension to create BM25 indexes, configure tokenizers, and build relevance-ranked queries directly within your database.

Hybrid search combines keyword and vector queries by integrating BM25 relevance ranking with pgvector semantic search, allowing you to perform hybrid keyword-plus-vector search across Postgres datasets for improved result accuracy.

Configure tokenizers and analyzers for Postgres full-text search by creating BM25 indexes with ParadeDB, enabling fuzzy matching, phrase queries, and custom tokenization rules to match Elasticsearch-quality text analysis.

Yes, you can use pgvector with ParadeDB for hybrid search entirely within Postgres, combining vector embeddings with BM25 keyword indexes to perform hybrid keyword-plus-vector search without external dependencies.

Postgres full-text search supports facets, aggregations, and snippet highlighting through ParadeDB's pg_search extension, providing query tuning, fuzzy matching, and advanced analytics comparable to Elasticsearch.

Limitations of using ParadeDB include version-specific feature availability and the need to fetch current documentation at runtime to verify support, as the extension relies on Postgres internals rather than a dedicated search infrastructure.