cozo

Project Overview

CozoDB is a general-purpose, transactional, relational database that uses Datalog for query, is embeddable but can also handle huge amounts of data and concurrency, and focuses on graph data and algorithms.

Key Features

• Datalog querying: Supports recursive Datalog and classical whole-graph algorithms
• Vector search: HNSW indices integrated seamlessly with Datalog
• Near-duplicate search: MinHash-LSH algorithm
• Full-text search: Built-in full-text search support
• JSON value support
• MVCC: Multi-Version Concurrency Control protects indices from corruption
• Multi-platform: Standalone service, embedded, Python/NodeJS/Clojure, mobile, browser WASM

Target Audience

• Developers building AI applications
• Data scientists who need graph databases and vector search
• Developers who need embeddable databases

Quick Start

1. Choose the platform that fits your needs (standalone service, Python/NodeJS/Clojure library, WASM, etc.)
2. Install CozoDB following the documentation
3. Define queries using Datalog

Core Advantages

• All-in-one: Relational, graph, vector, and full-text search in a single system
• Embeddable: Can be embedded as a library in your application
• High performance: Optimized HNSW implementation where basic vector operations become the limiting factor
• Memory efficient: Rust's RAII ensures memory is freed as soon as processing is done

System Requirements

• OS: Cross-platform (Linux, macOS, Windows)
• Architectures: x86_64, arm64
• Additional Dependencies: Varies by platform