# Digital Egypt Assistant | المساعد الرقمي لخدمات مصر 🇪🇬
🌐 **Live Demo:** digital-egypt-assistant.streamlit.app
An AI-powered **Arabic-language chatbot** that helps citizens navigate Egypt's digital government services (بوابة مصر الرقمية). Built with **Streamlit**, **LangChain**, **Groq LLMs**, and **RAG (Retrieval-Augmented Generation)** using a ChromaDB vector store and multilingual HuggingFace embeddings.
> اسألني عن أي خدمة رقمية متاحة على بوابة مصر الرقمية
---
## ✨ Features
- **Arabic-first & RTL Layout** — Native Right-to-Left Arabic UI with Egyptian identity styling and custom CSS layout.
- **RAG Architecture** — Retrieves relevant government service information from a local ChromaDB vector store before generating structured responses.
- **Groq LLMs Integration** — Ultra-fast response generation using active Groq models (`GPT-OSS 120B`, `GPT-OSS 20B`, `ALLaM 2 7B`).
- **Structured Service Cards & Suggestions** — Automatically extracts and formats service details (description, requirements, steps) and intelligently suggests **Similar Services**.
- **Session-aware Memory** — Conversation history is maintained across turns using `StreamlitChatMessageHistory`.
- **Conversation Summarization** — One-click sidebar button condenses the chat history into a short Arabic summary.
- **Quick-access Service Chips** — One-click buttons for popular services (vehicle fines, social insurance, insurance numbers, etc.).
- **Streaming Responses** — Real-time streaming answers chunk-by-chunk for a smooth user experience.
---
## 📊 RAG Model Evaluation / تقييم نموذج الـ RAG
Evaluation results using RAG Triad Metrics:
- **Context Relevance (صلة السياق):** `5.00 / 5.0` (100%)
- **Faithfulness (الدقة وعدم الهلوسة):** `5.00 / 5.0` (100%)
- **Answer Relevance (ملائمة الإجابة):** `5.00 / 5.0` (100%)
- **Overall Quality Score (التقييم الكلي للنظام):** `5.00 / 5.0` (100%)
Run evaluations anytime using:
```bash
python evaluate_rag.py
```
---
## 🛠️ Tech Stack
| Component | T …