MohamedSaeed130/Telecom-Egypt-mini-chatbot

# 🤖 Telecom-Egypt-mini-chatbot A production-ready RAG-powered intelligent chatbot for Telecom Egypt that answers customer questions using the official website as the primary knowledge base. ## Demo ## 📹 Project Demo ## Presentation ## Project Presentation ## Key Features - **Multi-lingual Support**: Handles Arabic (Modern Standard & Egyptian dialect) and English - **RAG-Powered**: Uses Retrieval Augmented Generation for accurate, grounded responses - **Document Upload**: Supports PDF, DOCX, TXT, HTML, and images (with OCR) - **Hybrid Retrieval**: Combined (Dense + Sparse[BM25]) search. - **Source Citations**: All answers include references to original sources - **Web-Based Interface**: Professional Streamlit chat interface ## Tech Stack - **Frontend**: Streamlit - **LLM**: Groq API (llama-3.3-70b-versatile) - **Vector DB**: Qdrant Cloud - **Embeddings**: HuggingFace (`intfloat/multilingual-e5-large`) - **Scraping**: Scrapy - **Document Processing**: `PyPDF2`, `python-docx`, `BeautifulSoup` ## Project Structure ``` Telecom-Egypt-mini-chatbot/ ├── src/ │ ├── data_chunking/ # Text chunking logic │ ├── data_extraction/ # Scrapy and Document processing │ │ └── data_extraction_scrapy/ # Scrapy project for web scraping │ │ └── data_extraction_processing/ # Document processing logic │ ├── data_indexer/ # Logic to index data into Qdrant │ ├── qdrant_vector_store_DB/ # Qdrant client manager │ ├── streamlit_app.py # Main Streamlit Application UI │ ├── main_setup.py # Script for setup and scraping pipeline │ ├── requirements.txt # Python dependencies │ └── qdrant_db/ # Local fallback for vector store ├── LICENSE # License file └── README.md # Project Documenta …