Skip to main content
Reference is where the exact ParadeDB syntax lives. Use these pages when you need operators, functions, index options, configuration settings, or the rules that determine whether ParadeDB can accelerate a query. Most ParadeDB features start with index design. Decide which columns and expressions belong in the ParadeDB index, including text fields, scalar fields used for filtering or sorting, JSON, arrays, composite fields, and vectors. Then choose the query pattern for your workload: full-text search, vector search, hybrid ranking, aggregates, facets, or joins. If you are new to ParadeDB, start with Getting Started and Architecture.

​
Build the Index

Create an Index

Create a ParadeDB index and choose the fields it should cover.

How the Index Works

Understand the inverted, columnar, vector, and segment structures inside the ParadeDB index.

Columnar Storage

Make fields available for filtering, sorting, grouping, and aggregates.

Specialized Indexing

Index expressions, JSON, arrays, composite fields, and vectors.

​
Query the Index

Full-Text / Lexical Search

Query text with match, term, phrase, proximity, fuzzy matching, regex, filters, sorting, relevance scores, and highlighting.

Vector / Semantic Search

Search vector values by semantic similarity and tune nearest-neighbor retrieval for larger datasets.

Hybrid Search

Combine BM25 and vector rankings into a single result set with reciprocal rank fusion.

​
Compose Results

Filtering and Sorting

Push filters and Top K ordering into the same index-backed query.

Aggregates and Facets

Compute SQL aggregates, buckets, histograms, metrics, and facets from the index.

Supported Join Types

Join search results with relational tables while keeping application data in Postgres.

​
API Surface

Operators and Functions

Look up query operators, vector operators, filter syntax, and aggregate helpers.

Configuration

Review the settings that control scans, aggregates, vectors, parallelism, and diagnostics.

SQL Functions

Look up callable SQL functions, signatures, compatibility names, and lower-level helpers.