salmalid/marrakech-muse-interface

# Morocco Travel Assistant ## Project Overview This is an NLP (Natural Language Processing) project developed as part of a Master's degree program, focusing on intelligent travel recommendations for Morocco. ### Current Implementation The chatbot currently uses **Google's Gemini API** to generate conversational responses and travel recommendations for Moroccan cities including: - 🍽️ Restaurants (Moroccan, Italian, French, Indian, Seafood, etc.) - 🏨 Hotels & Riads - ☕ Cafés - 🎭 Tourist Attractions **Note**: The system occasionally generates fictional recommendations when specific data is not available in the database. This is intentional behavior to maintain conversation flow and provide helpful suggestions. ### Future Goal The ultimate objective is to deploy a **custom prompt-engineered Mistral model** that has been specifically trained for Morocco travel recommendations. This migration is pending the availability of a **free cloud GPU with at least 16GB of memory**. **Why Mistral?** - Better control over responses - Reduced hallucination with proper prompt engineering - Cost-effective for production deployment - Fine-tuned specifically for Morocco tourism domain --- ## TEST IT : **URL**: marrakech-muse-interface-4r… ## Technologies Used This project is built with: - **Frontend**: React + TypeScript + Vite - **Styling**: Tailwind CSS + shadcn-ui - **AI/NLP**: Google Gemini API (temporary) → Mistral Model (working on deployment) - **Deployment**: Vercel (API) + Lovable (Frontend) --- ## Project Structure ``` ├── api/ │ └── chat.ts # Gemini API integration (will be replaced with Mistral) ├── src/ │ ├── components/ # React components │ ├── pages/ # Application pages │ └── ... └── README.md ``` ## Academic Notes **Current Status**: Functional with Gemini API **Target Status**: Mistral model deployment pending GPU availability : model available in my Kaggle account (chec …