Neo4j GraphRAG Book Example

This repository contains all code, data, and notebooks referenced in the Neo4j GraphRAG book. It provides a step-by-step guide to building a Retrieval-Augmented Generation (RAG) system using Neo4j, OpenAI, LangChain, and real-world financial filings.

---

📖 Project Overview

• Ingest and model company, filing, and holdings data as a knowledge graph in Neo4j
• Run retrieval-augmented queries and semantic search over filings and financial data
• Use Jupyter notebooks for hands-on experimentation

---

📂 Project Structure

• 01_PDF_Loader_for_Neo4j_GraphRAG.ipynb — Loads PDF and CSV data into Neo4j, builds the graph, and creates vector embeddings
• 02_Retreivers_notebook.ipynb — Interactive notebook for querying the graph, running RAG workflows, and experimenting with retrieval
• 03_Neo4j_agent_notebook.ipynb — (Optional) Agent-based exploration and advanced workflows
• data/ — Contains:
  • Asset_Manager_Holdings.csv, Company_Filings.csv: Sample data files
  • form10k-sample/: Example 10-K filings for graph ingestion
• Additional_Info/ — (Optional) Additional scripts and explained notebooks
• .env.sample — Template for required environment variables

---

🛠️ Setup Instructions

1. Clone the repository
2. Install dependencies

   pip install -r requirements.txt

3. Configure environment variables:
   • Copy .env.sample to .env and fill in:
     • OPENAI_API_KEY
     • NEO4J_URI, NEO4J_USERNAME, NEO4J_PASSWORD
4. Start Neo4j AuraDB (see book for connection details)
5. Run the loader notebook (01_PDF_Loader_for_Neo4j_GraphRAG.ipynb) to ingest data
6. Open 02_Retreivers_notebook.ipynb to run retrievals and explore the graph

---

📊 Data Files

• All sample data used in the book is in the data/ directory
• You may add or replace files in form10k-sample/ as needed for your experiments

---

🧹 Cleaning Up Unnecessary Files

• You may safely delete the following if not used in your workflow:
  • Additional_Info/__pycache__/
  • Any script or notebook in Additional_Info/ not referenced in the book
  • neo4j_importer_model_2025-05-08.json if not required
• Keep only the notebooks and data files you use in your book walkthrough

---

📝 Notes

• Never commit secrets — .env is in .gitignore
• Use Jupyter or VS Code for interactive exploration
• All notebooks are designed to be run step-by-step as described in the book

---

📚 Resources

• Neo4j AuraDB
• OpenAI API
• LangChain Docs
• Neo4j Python Driver

---

For questions or help, open an issue in this repo.