DuckDB Hybrid Document Search

Example Use Cases

• 1

  Index and search documentation written in Markdown with high precision.

• 2

  An efficient full-text and vector search hybrid solution.

• 3

  Seamless integration with AI models for document retrieval and serving.

Description

DuckDB Hybrid Document Search is designed for efficiently indexing internal documents written in Markdown. It combines full-text search with vector search to provide high-precision retrieval of information. The tool supports Japanese morphological analysis and uses a CrossEncoder for search result re-ranking. It integrates seamlessly with AI coding agents using MCP stdio server capabilities.