GraphWeaver

Graph RAG Go React Docker GitHub Action

GitHub Repository

Project Overview

GraphWeaver is a high-performance Graph RAG (Retrieval-Augmented Generation) knowledge engine. It leverages the power of “Entity-Relationship” modeling and “Semantic Vector” search to solve complex multi-hop reasoning challenges that traditional RAG systems often struggle with.

DEMO

demo

Key Features

  • Hybrid Retrieval Engine: Combines Vector Search (for semantic entry points) with Graph Diffusion (for logical relationship reasoning).
  • Containerized Deployment: Easy setup with Docker Compose for local development and testing.
  • High Performance: Backend implemented in Go (Golang) for efficient concurrency and low-latency processing.
  • Modern Knowledge Management: Automated entity and relationship extraction from unstructured data (PDF/Markdown).
  • Interactive UI: Sleek and responsive dashboard built with React, TypeScript, and Tailwind CSS.

System Architecture

diagram

Technology Stack

Backend & Core

  • Language: Go (Golang) 1.24+
  • Framework: Gin Gonic
  • LLM Integration: Gemini API

Databases & Storage

  • Graph Database: Neo4j (Core storage for relationship modeling)
  • Relational Database: PostgreSQL (For metadata management)
  • Vector Search: Qdrant (Planned for future hybrid retrieval implementation)

Frontend & Styling

  • Framework: React.js with TypeScript
  • Styling: Tailwind CSS

The front-end was primarily completed with the assistance of Vibe coding.

Infrastructure & DevOps

  • Containerization: Docker & Docker Compose
  • Orchestration: Kubernetes (K8s) (planning)
  • CI/CD: GitHub Actions

Development Status

This project is currently under active development, Currently, we’re deploying using Docker Compose; the Kubernetes solution might require more time for me to research, now focusing on optimizing graph data fetching and enhancing the chat service implementation.

← Back to Home