📌 Project Name: Complaint Processing with AI-Powered Classification System

🚀 Overview

Claimcat_1 is an AI-driven complaint categorization system designed to assist Bundesnetzagentur (BNetzA) in classifying consumer complaints efficiently. The system utilizes Natural Language Processing (NLP) models deployed on AWS SageMaker to automate the categorization process, reducing response time and workload for employees.

📡 Features

✅ Automated complaint categorization into Process, Type, and Content levels
✅ Multi-label classification for complaints covering multiple categories
✅ REST API-based integration with existing BNetzA workflows
✅ AWS Lambda & SageMaker-based deployment

🏗 System Architecture

• Frontend: Users submit complaints via web forms, emails, or letters.
• Backend: AWS Lambda processes the request and sends it to SageMaker for inference.
• AI Model: NLP-based classifier predicts complaint categories with confidence scores.
• API Response: Returns structured categorization results.

📡 API Endpoints

Endpoint	Description
/process-categorization	Classifies complaints as Strom, Gas, Sostinges.
/type-categorization	Identifies ****
/content-categorization	Categorizes .

📥 Example API Request

{
    "text": "I have issues with my electricity bill and connection."
}

📤 Example API Response

{
    "process_category": "Energy",
    "type_category": "Inquiry",
    "content_categories": ["Electricity", "Billing", "Network Connection"],
    "confidence_scores": [98.3, 95.5, 92.1]
}

🔧 Installation & Setup

1️⃣ Clone the repository

git clone https://github.com/srishtiguptaunimi/Claimcat_1.git
cd Claimcat_1

2️⃣ Install dependencies

pip install -r requirements.txt

3️⃣ Deploy on AWS Lambda

1. Set up an AWS account.
2. Create an IAM Role with permissions for SageMaker & Lambda.
3. Deploy Lambda using AWS CLI or the console.
4. Update API Gateway to expose endpoints.

🛠 Model Training & Deployment

• The NLP model is trained using consumer complaint data.
• Fine-tuned using SageMaker PyTorch models.
• Supports multi-class classification with confidence scores.
• Deployment: The model is hosted on AWS SageMaker Inference Endpoint.

🧪 Running Tests

• Unit Tests:

pytest tests/

• API Testing: Use Postman or curl:

curl -X POST "https://your-api-endpoint.com/process-categorization"      -H "Content-Type: application/json"      -d '{"text": "I have issues with my gas bill"}'

🤝 Contribution Guidelines

We welcome contributions! To contribute:

1. Fork this repository.
2. Create a new feature branch: git checkout -b feature-name
3. Commit changes: git commit -m 'Add new feature'
4. Push: git push origin feature-name
5. Create a Pull Request.

📜 License

This project is licensed under the MIT License.

📩 Contact & Support

For questions, reach out via GitHub Issues or email: [email protected]

🚀 Future Enhancements

• ✅ Improve multi-label classification for complex complaints.
• ✅ Build a FastAPI frontend for better integration.
• ✅ Add automated model retraining pipeline using AWS Lambda.

---

💡 Developed by: ClaimCat Team | Powered by AWS SageMaker, Bedrock & Lambda 🚀