Agentic Graph RAG (LangChain + LangGraph + Groq + Neo4j)

A generic, domain-agnostic knowledge graph extraction and RAG system that works with any PDF document. The system automatically analyzes document content, generates an appropriate schema, extracts entities and relationships, and builds a queryable knowledge graph.

Features

• 🔄 Fully Generic: Works with any PDF content - research papers, business docs, technical manuals, reports, etc.
• 🤖 Auto-Schema Generation: Automatically analyzes document content and creates appropriate entity/relationship schemas
• 📊 Smart Extraction: Uses LLM-powered agents to extract comprehensive entities and relationships
• 💾 Neo4j Integration: Stores extracted knowledge in a graph database for powerful queries
• 💬 Natural Language Chat: Ask questions about your documents using natural language
• 🔍 Direct Cypher Queries: Run custom graph queries for advanced analysis

Quick Start

1. Setup Environment

# Install dependencies
pip install -r requirements.txt

# Copy and configure .env file
cp .env.example .env
# Edit .env with your API keys and Neo4j credentials

2. Run the Pipeline

python main.py

Choose option 1, upload any PDF, and let the system automatically:

• Analyze the document content
• Generate an appropriate schema
• Extract entities and relationships
• Store everything in Neo4j

3. Query Your Knowledge Graph

Use option 2 to chat with your knowledge graph using natural language!

Docker Compose (Optional)

docker compose up --build -d
# Pipeline (app), HIL UI http://localhost:8501, Jupyter http://localhost:8888

Notebooks (Jupyter)

Open http://localhost:8888 and try:

• notebooks/pagerank.ipynb
• notebooks/community.ipynb
• notebooks/link_prediction.ipynb
• notebooks/link_prediction_pipeline.ipynb

Bloom

See docs/bloom.md.