bekalu73/ethiopian-constitution-rag

AI-powered RAG assistant for the Constitution of Ethiopia using LLMs, semantic search, and vector embeddings. # Ethiopian Constitution RAG A Retrieval-Augmented Generation (RAG) application that lets you ask questions about the Ethiopian Constitution and get accurate, context-aware answers with page references. ## Architecture ``` PDF → Chunking → HuggingFace Embeddings → Qdrant Vector DB ↓ User Query → Similarity Search → Context + Gemini LLM → Answer ``` **Stack:** - **Embeddings:** `BAAI/bge-small-en-v1.5` (local, free) - **Vector DB:** Qdrant (Docker) - **LLM:** Google Gemini via OpenAI-compatible API - **Framework:** LangChain ## Prerequisites - Python 3.10+ - Docker - A Google AI API key (get one here) ## Setup **1. Clone the repo** ```bash git clone github.com cd ethiopian-constitution-rag ``` **2. Start Qdrant** ```bash docker compose up -d ``` **3. Install dependencies** ```bash python -m venv venv source venv/bin/activate pip install -r requirements.txt ``` **4. Configure environment** ```bash cp .env.example .env ``` Add your Gemini API key to `.env`: ``` GEMINI_API_KEY=your_api_key_here ``` ## Usage **Step 1 — Index the constitution** ```bash python index.py ``` This loads the PDF, splits it into chunks, generates embeddings, and stores them in Qdrant. **Step 2 — Start chatting** ```bash python chat.py ``` Type your question and get an answer with page references. ## Project Structure ``` ├── chat.py # Interactive Q&A script ├── index.py # PDF indexing script ├── Ethiopia_Constitution.pdf # Source document ├── docker-compose.yml # Qdrant service ├── requirements.txt # Python dependencies └── .gitignore ``` ## How It Works 1. **Indexing** (`index.py`) — Loads the Ethiopian Constitution PDF, splits it into 1000-character chunks with 200-character overlap, generates embeddings using HuggingFace, and stores them in Qdrant. 2. **Querying** (`chat.py`) — Takes a user question, performs similarity search against the vector store, ret …