EdgeQuake

EdgeQuake is a high-performance Graph-RAG framework developed in Rust, designed to transform documents into intelligent knowledge graphs for superior information retrieval and content generation.

Key Features

• Intelligent Knowledge Graphs: Enable advanced entity extraction and relationship mapping, capturing the structural relationships between concepts rather than relying solely on semantic similarity. This comprehensive understanding enhances lookup capabilities beyond simple keyword matching.

• Flexible Query Modes: Utilizes six different query modes optimized for various question types, allowing for efficient querying that ranges from fast vector searches to complex graph traversals. Each mode caters to different scenarios, such as specific relationship inquiries or thematic explorations.

• Optimized Performance: Built with a Tokio-based asynchronous architecture, EdgeQuake efficiently handles thousands of concurrent requests using zero-copy operations for enhanced memory management.

• Advanced Document Processing: Features robust PDF processing capabilities, including table detection, OCR functionalities, and multi-column layout handling to effectively manage a variety of document types.

• Production-Ready Architecture: Includes a comprehensive REST API compliant with OpenAPI 3.0, ensuring smooth integration and operational capabilities, along with health checks and multi-tenant workspace isolation.

• Modern User Interface: Offers a React-based frontend that provides real-time streaming of results, interactive visualizations, and a user-friendly configuration interface.

Performance Metrics

EdgeQuake demonstrates significant performance improvements over traditional RAG systems, as showcased in the following benchmarks:

Metric	EdgeQuake	Traditional RAG	Improvement
Entity Extraction	2-3x faster	Baseline	3x
Hybrid Query Latency	< 200ms	~1000ms	5x faster
Document Processing Time	25s (10k tokens)	~60s	2.4x faster
Concurrent Users	1000+	~100	10x
Memory Usage (per doc)	2MB	~8MB	4x better

How EdgeQuake Works

EdgeQuake implements the LightRAG algorithm, which leverages knowledge graph extraction during document indexing, followed by graph traversal during querying. The indexing process involves:

1. Chunking documents into manageable segments.
2. Extracting entities and relationships using language models.
3. Normalizing to reduce duplicates.
4. Embedding for vector representations.
5. Storing in PostgreSQL, integrating graph and vector storage.

Components

• REST API: Facilitates interaction through API endpoints with extensive documentation and support for real-time streaming responses.
• Frontend: Built using React, providing a rich user experience with drag-and-drop capabilities for document uploads and engaging graph visualizations.

Documentation and Community

Comprehensive documentation is available, detailing setup, feature usage, and architecture. Community support is encouraged through GitHub issues and discussions for user collaboration and feedback.

For more information and to explore EdgeQuake's capabilities, visit the repository. With its innovative approach to knowledge graph construction and retrieval, EdgeQuake stands as a pivotal solution for organizations seeking efficient document management and intelligent information retrieval.