Graphy: GraphRAG with LangGraph

Graphy is a powerful Graph Retrieval-Augmented Generation (GraphRAG) application built with LangGraph, Groq, and Neo4j. It allows users to upload PDF documents, extract structured knowledge into a graph database, and perform complex queries using natural language.

🚀 Features

• PDF Knowledge Extraction: Upload PDF documents and automatically extract entities and relationships.
• GraphRAG Engine: Combines the power of Knowledge Graphs with LLMs for more accurate and context-aware answers.
• Natural Language Querying: Ask questions in plain English; the app converts them into Cypher queries to retrieve data from Neo4j.
• Powered by Groq: Utilizes the lightning-fast inference of Groq with support for multiple models (Llama 3.3, Llama 3.1, Mixtral, Gemma) to ensure reliability and performance.
• LangGraph Orchestration: Uses a stateful graph workflow to manage the query generation, execution, and response formatting pipeline.

🛠️ Tech Stack

• Frontend: Streamlit
• LLM: Llama 3.3, Llama 3.1, Mixtral, Gemma 2 (via Groq API)
• Graph Database: Neo4j
• Orchestration: LangGraph
• Framework: LangChain
• Embeddings: HuggingFace (sentence-transformers/all-MiniLM-L6-v2)
• PDF Processing: PyMuPDF

📋 Prerequisites

Before running the application, ensure you have the following:

1. Python 3.9+ installed.
2. A Neo4j Database (AuraDB Free tier works great).
3. A Groq API Key.

⚙️ Installation

1. Clone the repository:

   git clone <repository-url>
   cd <repository-folder>

2. Install dependencies:

   pip install streamlit python-dotenv langchain-community langchain-core langchain-groq langchain-huggingface langchain-neo4j langchain-experimental langgraph pymupdf neo4j

3. Set up Environment Variables: Create a .env file in the root directory and add your credentials:

   GROQ_API_KEY=your_groq_api_key
   NEO4J_URI=neo4j+s://your-instance.databases.neo4j.io
   NEO4J_USERNAME=neo4j
   NEO4J_PASSWORD=your_password

🏃‍♂️ Usage

1. Run the Streamlit app:

   python -m streamlit run main.py

2. Configure & Connect:

   • Enter your Groq API Key and Neo4j credentials in the sidebar (or use .env).
   • Select your preferred LLM Model (e.g., Llama 3.3, Mixtral) from the dropdown to handle rate limits or optimize performance.
   • Click Connect & Initialize.

3. Upload a PDF:

   • Use the file uploader to select a PDF document.
   • The app will process the file, split the text, and extract a knowledge graph.

4. Ask Questions:

   • Type your question in the input box (e.g., "What symptoms does the patient have?").
   • The app will generate a Cypher query, execute it against the graph, and provide a natural language answer.

🧠 How It Works

1. Ingestion: The PDF is loaded using PyMuPDFLoader and split into chunks.
2. Extraction: LLMGraphTransformer (powered by the selected Groq model) analyzes the text to identify nodes (e.g., Patient, Disease, Medication) and relationships.
3. Storage: The extracted graph data is stored in Neo4j.
4. Retrieval (GraphRAG):
   • Generate Cypher: The LLM translates the user's question into a Cypher query based on the graph schema.
   • Execute: The query is run against the Neo4j database.
   • Answer: The results are formatted and synthesized into a final answer by the LLM.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.