Skip to content

Text Search Features

AkasicDB text search provides tantivy-based full-text indexing and BM25 ranking as PostgreSQL indexes. Without a separate search infrastructure, you create an index with CREATE INDEX ... USING texton and search with the @@ operator and a query DSL. Multilingual tokenization including Korean morphological analysis, sorting and pagination, highlighting, and aggregation all run inside a single database.

Topics

  1. Indexing Covers USING texton index syntax, whole-row / multi-column indexes, field mapping (mapping), and fast fields.

  2. Tokenization Covers analyzers and tokenizers, Korean morphological analysis with lindera ko-dic, per-field and user dictionaries, and the dictionary path allowlist.

  3. Querying Covers the @@ search operator and the texton.query DSL, BM25 scoring (akasicdb.text_score()), sorting and cursor pagination (akasicdb.text_search()), highlighting, and aggregation.

  4. Operations Covers batch commit and akasicdb.text_flush(), reindexing, CDC ingestion, replica segment shipping, observability SRFs, and runtime settings (GUCs).