DramaW0r/tunisia-archaeological-chatbot

🏛️ AI-powered chatbot exploring Tunisian archaeological sites using RAG (Retrieval-Augmented Generation) with Llama3, ChromaDB & Streamlit. Local-first, conversation-aware, with source citations for heritage discovery. # Tunisia Archaeological Chatbot An intelligent RAG-based chatbot for exploring Tunisia's archaeological heritage through natural language conversations. ## Overview This system combines semantic search, vector databases, and large language models to provide accurate responses about Tunisian archaeological sites with full source attribution. **Key Features:** - Semantic search using ChromaDB and Sentence Transformers - Context-aware conversations with Llama3 - Automatic source citation with relevance scoring - Persistent chat history with SQLite - Local-first deployment (no external APIs) ## Features ### Core Functionality - **RAG Pipeline**: Vector embeddings with Sentence Transformers and semantic search via ChromaDB - **Conversational AI**: Multi-turn dialogues with context awareness using Llama3 - **Source Attribution**: Automatic citations with relevance scores and metadata - **Persistence**: SQLite-backed chat history with search and analytics ### User Interface - Clean Streamlit web interface - Real-time response streaming - Conversation management (create, load, delete, search) - System status monitoring - Privacy-first local deployment ## Architecture ### System Components ``` User Interface (Streamlit) ↓ RAG Engine ↓ ↓ ChromaDB Ollama/Llama3 ↓ SQLite Database ``` ### Technology Stack - **Web Framework**: Streamlit - **Vector Database**: ChromaDB - **Embeddings**: Sentence Transformers (all-MiniLM-L6-v2) - **LLM**: Llama3 via Ollama - **Database**: SQLite3 - **Language**: Python 3.8+ ## Quick Start ### Prerequisites - Python 3.8+ - Ollama (Installation) - 4GB RAM minimum ### Installation **1. Clone and setup environment** ```bash git clone github.com cd tunisia-archaeological-chatbot python -m venv venv source venv/bin/activate # Windows: venv\Scripts\activate pip install -r requirements.txt ``` **2. Configure Ollama** ```bash ollama pull llama3 ollama serve # Keep runni …