Shu'ur is an AI-powered platform that automatically analyzes, classifies, and visualizes Arabic customer support conversations (including Palestinian dialect) to help organizations instantly understand customer sentiment, spot dissatisfaction, and improve service quality.

• Live Demo: Watch a demo here

• Arabic Sentiment Analysis: Detects positive, negative, and neutral sentiment in Arabic (including dialects).
• Real-Time Dashboard: Visualizes sentiment trends, agent performance, and common issues.
• Automatic Tagging: Highlights dissatisfied customer messages for rapid intervention.
• Comprehensive Metrics: Accuracy, Precision, Recall, F1-Score, MAE, RMSE, and more.
• Real Time Chatting: Designed for live chat, and web-based customer support systems.
• Secure API Integration: Uses OpenRouter LLM APIs with API key authentication.

• Live Demo: Demo Video

1. API Authentication
   Secure connection to OpenRouter LLM APIs with your API key.

2. Prompt Creation
   Each customer message is formatted and sent to the LLM with context.

3. Model Inference
   The LLM returns a sentiment score and explanation.

4. Data Analysis & Visualization
   Results are aggregated, analyzed, and visualized in real-time dashboards.

• Python (backend, AI model interaction)
• Jupyter Notebook (data analysis, prototyping)
• TypeScript (dashboard/web frontend)
• OpenRouter API (LLM inference)
• Mistral AI (pre-trained models)
• Matplotlib, Seaborn (visualizations)
• Pandas, scikit-learn (data processing & metrics)

• ai-model/ — Python backend for sentiment analysis
• data-analysis/ — Jupyter Notebooks for data exploration and evaluation
• web/ — TypeScript frontend for the dashboard

1. Clone the repository

   git clone https://github.com/YOUR-TEAM-ORG/ai-model.git
   

2. Install dependencies

   cd ai-model
   pip install -r requirements.txt
   

3. Set your OpenRouter API key

   export OPENROUTER_API_KEY=your_api_key_here
   

4. Run the backend

   python evaluate_model.py
   

5. Start the frontend dashboard

   cd ../web
   npm install
   npm start
   

• Production Deployment: Integrate with Bank of Palestine’s infrastructure
• Dialect Expansion: Support for more Arabic dialects
• Real-time Alerts: Live dashboards and automated escalation
• Predictive Analytics: Customer satisfaction forecasting
• Multi-channel Support: Voice, email, social media integration

Palestine Launchpad 4th Graduate Hackathon - Team20-P3

Contact: [email protected]