Repository navigation
v0.1.0
v0.1.0
Release Date: 2025-08-03
Initial release of AGraph - A comprehensive toolkit for building, storing and querying knowledge graphs.
π Features
Core Components
- Entity System - Robust entity data structures with type definitions and operations
- Relations Framework - Comprehensive relationship modeling and management
- Knowledge Graph Core - Central graph data structure for organizing and querying knowledge
Document Processing
- Multi-format Support - Process PDF, Word, Text, HTML, Spreadsheet, JSON, and Image files
- Factory Pattern Architecture - Extensible processor factory for adding new document types
- Batch Processing - Efficient handling of multiple documents and folders
Graph Builders
- StandardGraphBuilder - Rule-based entity and relation extraction
- LightRAGGraphBuilder - LLM-powered graph building using the LightRAG framework
- MultiSourceGraphBuilder - Integration of multiple data sources into unified graphs
Storage Backends
- JsonStorage - File-based JSON storage for lightweight deployments
- Neo4jStorage - Neo4j graph database integration for enterprise applications
- Abstract Storage Interface - Easy extension for custom storage backends
Extractors
- Entity Extractors - Extract entities from text and database sources
- Relation Extractors - Identify and extract relationships between entities
- Extensible Architecture - Base interfaces for implementing custom extractors
Embeddings
- Graph Embeddings - Node2Vec and TransE algorithms for graph representation learning
- Entity Similarity - Compute semantic similarity between entities
- Vector Operations - Support for downstream ML tasks
LightRAG Integration
- LLM-powered Extraction - Automatic entity/relation extraction using language models
- GraphML Export - Standard format support for visualization tools
- Multiple Search Modes - Naive, local, global, and hybrid search capabilities
- Incremental Updates - Add documents to existing graphs without rebuilding
π§ Developer Experience
API Design
- Convenience Functions - One-line graph creation with
quick_build_graph() - Factory Methods - Simple creation of builders and storage instances
- Type Safety - Full type annotations for better IDE support
Testing & Quality
- Comprehensive Test Suite - 95%+ code coverage across all modules
- Quality Tools - Black, isort, mypy, flake8, bandit integration
- CI/CD Ready - Pre-commit hooks and automated quality checks
Documentation
- Usage Examples - Complete examples for common workflows
- API Documentation - Detailed documentation for all public APIs
- Architecture Guide - Clear explanation of design patterns and extensibility
π¦ Installation
pip install agraphOptional Dependencies
# Development tools
pip install agraph[dev]
# Documentation tools
pip install agraph[docs]
# All dependencies
pip install agraph[all]π Quick Start
from agraph import quick_build_graph, create_json_storage
# Build a knowledge graph
graph = quick_build_graph(
texts=["Knowledge graphs represent structured information"],
graph_name="my_graph"
)
# Save to JSON storage
storage = create_json_storage("./graphs")
storage.save_graph(graph)ποΈ Architecture
- Factory Pattern - Consistent object creation across the framework
- Strategy Pattern - Multiple algorithms for graph building and processing
- Plugin Architecture - Easy extension with custom components
- Abstract Base Classes - Well-defined interfaces for all major components
π Requirements
- Python - 3.10+ required
- Core Dependencies - BeautifulSoup4, LightRAG, OpenAI, PyPDF, Neo4j driver
- Optional Dependencies - Development and documentation tools
π‘οΈ Stability
This is the initial release (v0.1.0) of AGraph. While the core functionality is stable and well-tested, the API may evolve based on user feedback. We recommend pinning to specific versions in production environments.