Nigerian Constitution AI Chatbot 🇳🇬

A Retrieval-Augmented Generation (RAG) powered chatbot that answers questions about the Nigerian Constitution with legal citations.

Features

• 📄 PDF Document Processing with Metadata Extraction
• 🧠 RAG Implementation with FAISS Vector Store
• ⚖️ Legal Citation Tracking (Articles & Sections)
• ❓ Ambiguity Detection and Query Validation
• 🤖 Support for Both OpenAI and Local LLMs (Llama.cpp)
• 🌐 Gradio Web Interface

Installation

1. Clone Repository

git clone https://github.com/Joshuaatanu/law_rag.git
cd law_rag

2. Set Up Virtual Environment

python -m venv constitution-env
source constitution-env/bin/activate  # Linux/Mac
# constitution-env\Scripts\activate  # Windows

3. Install Dependencies

pip install -r requirements.txt

4. Download Constitution PDF

• Place nigerian_constitution.pdf in data/ directory
• Recommended source: Nigeria Law

Configuration

1. For OpenAI Users

echo "OPENAI_API_KEY=your_api_key_here" > .env

2. For Local LLM Users

wget https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF/resolve/main/mistral-7b-instruct-v0.1.Q4_K_M.gguf -O models/

Usage

1. Run the Application

python src/chatbot.py

2. Access Web Interface

• Local URL: http://localhost:7860
• Gradio Share URL will be displayed in console

3. Example Questions

• "What does Article 21 say about cultural rights?"
• "Explain the impeachment process in Section 143"
• "What are the fundamental rights in Chapter IV?"

Deployment

Option 1: Docker

docker build -t constitution-chatbot .
docker run -p 7860:7860 constitution-chatbot

Option 2: Hugging Face Spaces

1. Create new Space with Gradio SDK
2. Upload:
   • src/
   • data/
   • requirements.txt
3. Set OpenAI API key in Space secrets

Project Structure

├── data/
│   └── nigerian_constitution.pdf
├── models/
│   └── mistral-7b-instruct-v0.1.Q4_K_M.gguf
├── src/
│   ├── chatbot.py
│   ├── data_processing.py
│   └── config.py
├── requirements.txt
├── .env.example
└── README.md

Troubleshooting

Common Issues:

• API Key Errors: Verify .env file exists with correct credentials
• Model Not Found: Check model path in chatbot.py
• PDF Extraction Issues: Ensure PDF is not scanned/image-based

Performance Tips:

• Use n_ctx=4096 for longer context with local models
• Reduce chunk_size in config.py for better relevance
• Enable GPU acceleration for local models

Disclaimer

⚠️ This application provides general constitutional information and should not be considered legal advice. Always consult qualified legal professionals for official matters.

License

MIT License - See LICENSE for details.

Contributing

Contributions are welcome! Please open an issue or PR for:

• Additional legal provisions
• Improved citation tracking
• Local language support (Hausa/Yoruba/Igbo)

Acknowledgments

• LangChain Framework
• Gradio
• Nigerian Law Repository
• Mistral-7B model by TheBloke