Kuzu

Stale
GitHub C++ MIT

Description

An embedded property graph database built for speed with built-in vector search and full-text search, implementing Cypher query language for knowledge graph construction and AI agent structured knowledge retrieval.

Key Features

  • Embedded property graph database: runs directly embedded in applications without server processes, supports Cypher query language and flexible property graph data model
  • Native vector search and full-text search: built-in vector index and full-text search (FTS) capabilities for semantic retrieval without external vector databases
  • Columnar storage and high-performance queries: columnar disk storage with CSR adjacency list indices, vectorized query processor, and multi-core parallelism for large-scale graph analytics
  • WebAssembly bindings: execute graph queries securely and quickly in the browser via Wasm for client-side graph computation
  • ACID transactions and extension framework: supports serializable ACID transactions for data consistency, with extensible framework for dynamically loading algorithms, JSON and other modules
  • Multi-language client support: provides client bindings for Python, Node.js, Rust, Java and other languages

Use Cases

💡 AI agent knowledge graph construction: transform unstructured documents into property graphs, enabling complex entity relationship reasoning and knowledge retrieval through Cypher queries
💡 Graph-structured retrieval in RAG pipelines: combine vector search and graph traversal to provide structured knowledge retrieval for RAG systems, surpassing semantic limitations of pure vector retrieval
💡 Embedded graph analysis applications: embed graph database capabilities in mobile apps, desktop apps or IoT devices for localized social network analysis and recommendation systems
💡 Real-time graph data query services: leverage high-performance columnar storage and parallel query engines to provide real-time query services for applications requiring low-latency graph traversal

Tags

database vector-database rag embedding graph cpp

Categories

📚 RAG Tools

Quick Start

Install: pip install kuzu; create database: import kuzu; db = kuzu.Database('my_db'); conn = kuzu.Connection(db); create table and import: conn.execute('CREATE NODE TABLE Person(name STRING, age INT64, PRIMARY KEY(name))'); query: conn.execute('MATCH (p:Person) RETURN p.name, p.age')
Visit GitHub Visit Website

Related Projects