Inventory Graph Chatbot

Overview

This project provides an advanced, LLM-powered chatbot for querying and analyzing Mahindra & Mahindra's inventory data using a Neo4j knowledge graph. It leverages LangChain, Azure OpenAI, and custom prompt engineering to enable natural language Q&A, Cypher query generation, and data normalization.

Features

• Natural Language Chatbot: Ask inventory-related questions in plain English.
• Automated Cypher Generation: Converts user queries to Cypher for Neo4j.
• Entity Normalization: Maps user terms to canonical KG forms.
• Graph Data Transformation: Converts tabular/text data into graph documents.
• Extensible Prompt Templates: Easily adapt prompts for new use cases.
• Professional Notebook Workflows: Includes Jupyter notebooks for experimentation and development.

Project Structure

app.py                      # (Optional) App entry point
create_chatbot.ipynb        # Main notebook for chatbot logic
create_kg_new.ipynb         # (Optional) KG creation notebook
main.py                     # Main Python script for end-to-end execution
parse_excel.ipynb           # Data parsing notebook
prompts.py                  # All prompt templates
requirements.txt            # Python dependencies
visualization.png           # Project architecture/graph visualization
__pycache__/                # Python cache files
 data/                      # Inventory data files
 docs/                      # Documentation and schema files

Setup Instructions

1. Clone the Repository

git clone https://github.com/HiBorn4/graphRag.git
cd MahindraMahindra/archive/graphrag

2. Install Python Dependencies

It is recommended to use a virtual environment:

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

3. Configure Environment Variables

Create a .env file in the project root with the following variables:

NEO4J_URI=neo4j://localhost:7687
NEO4J_USER=neo4j
NEO4J_PASSWORD=your_neo4j_password
AZURE_DEPLOYMENT=your_azure_deployment
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint
AZURE_OPENAI_VERSION=your_azure_openai_version

4. Prepare Neo4j Database

• Install Neo4j Community or Aura and start the server.
• Load your inventory data as per the schema in docs/kg_schema_overview.md or docs/kg_schema_overview3.md.
• Update the .env file with your Neo4j credentials.

5. Run the Chatbot

Option 1: Jupyter Notebook

• Open create_chatbot.ipynb or main.py in VS Code or Jupyter Lab.
• Run cells step by step to interact with the chatbot and see Cypher queries/results.

Option 2: Python Script

python main.py

6. Customization

• Prompts: Edit prompts.py to adjust normalization or Cypher generation logic.
• Data: Place new inventory data in the data/ folder and update notebooks/scripts as needed.
• Visualization: Update visualization.png to reflect your graph structure.

Notebooks

• create_chatbot.ipynb: End-to-end chatbot logic, entity normalization, Cypher generation, and result analysis.
• graph_transformer.ipynb: Demonstrates transforming tabular/text data into graph documents using LLMs.
• parse_excel.ipynb: Utilities for parsing and preparing inventory data.

Example Usage

• Ask: "What is the difference in Stock in trade count for the company 133-FES from FY24 to FY23?"
• The system normalizes entities, generates Cypher, queries Neo4j, and returns a human-readable answer.

Troubleshooting

• Neo4j Connection Errors: Ensure Neo4j is running and credentials are correct in .env.
• Azure/OpenAI Errors: Verify your API keys and endpoint URLs.
• Missing Data: Check that your data files are present in the data/ directory and loaded into Neo4j.

License

This project is for internal use at Mahindra & Mahindra. For external use, please contact the project owner.

Contact

For questions or support, contact the project maintainer or your internal IT team.