Graph RAG Application

Medical knowledge graph and RAG system for processing pharmaceutical instructions and medical documents.

Project Structure

2025_graph_rag/
├── app/                      # Main application package
│   ├── clients/              # Client management
│   │   ├── http_client.py    # HTTP client configuration
│   │   └── llm_client.py     # LLM client management
│   ├── config.py             # Application configuration
│   ├── dependencies.py       # Shared dependencies
│   ├── models/               # Data models
│   │   └── prompts.py        # LLM prompt templates
│   ├── services/             # Business logic services
│   │   ├── document_service.py   # Document processing
│   │   ├── entity_service.py     # Entity extraction
│   │   ├── qa_service.py         # Q&A functionality
│   │   └── relation_service.py   # Relation extraction
│   ├── ui/                   # User interface
│   │   └── gradio_app.py     # Gradio web interface
│   └── utils/                # Utility functions
│       └── text_utils.py     # Text processing utilities
├── qdrant_manager.py         # Qdrant vector database manager
├── graph_builder.py          # Graph visualization builder
├── main.py                   # Application entry point
├── docker-compose.yml        # Docker services configuration
├── Dockerfile                # Application Docker image
└── pyproject.toml           # Project dependencies

Setup

1. Start Qdrant Vector Database

docker compose up -d

2. Install uv Package Manager

Linux/Mac:

wget -qO- https://astral.sh/uv/install.sh | sh

Windows:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

More info: https://docs.astral.sh/uv/getting-started/installation/#standalone-installer

3. Install Dependencies

uv sync

4. Configure Environment

Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key_here

# Optional: Proxy configuration
# PROXY_URL=http://your-proxy:port

# Optional: Qdrant configuration
# QDRANT_URL=http://localhost:6333

# Optional: Processing configuration
# CHUNK_SIZE=800
# CHUNK_OVERLAP=150
# MAX_WORKERS=4

5. Run the Application

python3 main.py

The Gradio interface will be available at http://localhost:7860

Docker Deployment

To run the entire stack with Docker:

docker compose up -d

This will start:

• Qdrant vector database (ports 6333, 6334)
• Gradio application (port 7860)

Features

• Document Processing: Upload and process PDF and Markdown medical documents
• Entity Extraction: Automatically extract medical entities (diseases, drugs, symptoms, etc.)
• Relation Extraction: Identify relationships between medical entities
• Vector Search: Semantic search using Qdrant vector database
• Knowledge Graph: Interactive visualization of extracted entities and relations
• Q&A System: Answer questions based on processed documents
• Session Management: Organize documents into isolated sessions