AndrewMbugua/KenyaLawQuery-DevFest-Kisii-2025

# Kenyan Law Document Query System A LangChain-based system for querying Kenyan law documents and court cases using Python and Ollama for local LLM inference. ## Overview This system allows you to: - Load and process PDF files containing Kenyan law documents and court cases - Create a searchable vector database using FAISS - Query the documents using natural language with Ollama's local LLM - Get answers with citations to specific source documents and page numbers ## Prerequisites ### 1. Ollama Installation You must have Ollama installed on your Linux system. Install it with: ```bash curl -fsSL ollama.com | sh ``` ### 2. Download Ollama Models Download the required models: ```bash # Download the LLM model (choose one) ollama pull llama3.2 # Recommended # OR ollama pull mistral # OR ollama pull llama2 # Download the embedding model ollama pull nomic-embed-text ``` ### 3. Start Ollama Service Make sure Ollama is running: ```bash ollama serve ``` This should start Ollama on `localhost` ## Setup ### 1. Environment Configuration Copy the example environment file: ```bash cp .env.example .env ``` Edit `.env` to configure your settings (optional, defaults should work): ```env OLLAMA_BASE_URL=localhost OLLAMA_MODEL=llama3.2 OLLAMA_EMBEDDING_MODEL=nomic-embed-text PDF_DIRECTORY=./kenyan_law_pdfs CHUNK_SIZE=1000 CHUNK_OVERLAP=200 VECTOR_STORE_PATH=./faiss_index ``` ### 2. Add PDF Documents Place your Kenyan law PDF files in the `kenyan_law_pdfs` directory: ```bash cp /path/to/your/pdfs/*.pdf kenyan_law_pdfs/ ``` The system will recursively process all PDF files in this directory. ## Usage ### Step 1: Index Documents First, you need to index all the PDF documents to create the vector database: ```bash python main.py index ``` This will: - Load all PDF files from `kenyan_law_pdfs/` - Split documents into chunks - Generate embeddings using Ollama - Create and save a FAISS vector store **Note:** Thi …