Skip to content

REPOSITORY_SUMMARY.md

Latest commit

 

History

History
115 lines (98 loc) · 4.43 KB

REPOSITORY_SUMMARY.md

File metadata and controls

115 lines (98 loc) · 4.43 KB
Raw

LlamaGraph Repository Summary

Overview

LlamaGraph is a llama-themed knowledge graph construction tool that extracts entities and relationships from text and builds a queryable knowledge graph. It enables users to process unstructured text data, extract meaningful information, and represent it in a structured format that can be queried and analyzed.

Core Features

  • Entity extraction using SpaCy with optional MLX acceleration
  • Relationship extraction between entities
  • Knowledge graph construction using NetworkX
  • Interactive query capabilities
  • Graph visualization and analysis tools
  • Command-line interface with llama-themed UI elements
  • REST API for programmatic access

Technical Architecture

LlamaGraph follows a modular architecture with the following core components:

  1. Entity Extraction: Identifies named entities in text using SpaCy and custom extractors with MLX acceleration.
  2. Relation Extraction: Identifies relationships between entities using pattern matching and dependency parsing.
  3. Knowledge Graph: Builds and manages a graph structure to represent entities and their relationships.
  4. Query Engine: Provides a flexible interface for querying the knowledge graph.
  5. Terminal UI: Offers a colorful, interactive command-line interface for users.
  6. Cache System: Optimizes performance by caching extraction results.
  7. API Server: Exposes functionality through a REST API.

Technical Implementation

  • Written in Python 3.9+
  • Uses SpaCy for NLP processing
  • NetworkX for graph representation
  • MLX for performance optimization on Apple Silicon
  • Rich, textual, and prompt_toolkit for terminal UI
  • FastAPI for the API server (optional dependency)

Directory Structure

llamagraph/
├── llamagraph/             # Main package directory
│   ├── __init__.py         # Package initialization
│   ├── cli.py              # Command-line interface
│   ├── config.py           # Configuration management
│   ├── extractor/          # Entity and relation extraction
│   ├── graph/              # Knowledge graph and query engine
│   ├── ui/                 # Terminal UI components
│   ├── utils/              # Utility functions and classes
│   ├── api/                # API components
│   └── server/             # Server implementation
├── tests/                  # Test suite
├── examples/               # Example usage scripts
├── docs/                   # Documentation
├── scripts/                # Utility scripts
└── setup.py                # Package setup

Usage Examples

LlamaGraph can be used in several ways:

  1. Command-line Interface:

    llamagraph process -f input.txt -o graph.json

  2. Python API:

    from llamagraph.config import LlamaGraphConfig
from llamagraph.extractor.entity_extractor import EntityExtractor
from llamagraph.extractor.relation_extractor import RelationExtractor
from llamagraph.graph.knowledge_graph import KnowledgeGraph

# Initialize components
config = LlamaGraphConfig()
entity_extractor = EntityExtractor(config)
relation_extractor = RelationExtractor(config)
kg = KnowledgeGraph()

# Process text
text = "Apple was founded by Steve Jobs in 1976."
entities = entity_extractor.extract(text)
relations = relation_extractor.extract(text, entities)

# Build knowledge graph
for entity in entities:
    kg.add_entity(entity)
for relation in relations:
    kg.add_relation(relation)

# Query the graph
apple = kg.get_entity("Apple")
relations = kg.get_relations("Apple")

  3. REST API:

    llamagraph serve

    Then send HTTP requests to interact with the API.

Dependencies

  • Core: spacy, networkx, rich, click
  • Acceleration: mlx (on Apple Silicon)
  • Visualization: matplotlib, pyvis, plotly (optional)
  • API: fastapi, uvicorn (optional)

Known Limitations

  • MLX acceleration only works on Apple Silicon devices
  • Some complex relationships may not be correctly extracted
  • Performance may degrade with very large texts or complex graphs
  • Visualization tools require additional dependencies

Future Work

  • Improved entity coreference resolution
  • Enhanced relation extraction using transformer models
  • More sophisticated query capabilities
  • Graph embedding for semantic similarity queries
  • Integration with vector databases
  • Web-based visualization interface