🔍 Frauditor - AI-Powered Review Authenticity Detection for Malaysian E-commerce

Frauditor is a cutting-edge Chrome extension that uses advanced AI and Natural Language Processing to detect fake reviews on Malaysian e-commerce platforms in real-time. Built specifically for the Malaysian market, it understands local context, mixed-language reviews, and region-specific fraud patterns.

Fake Review Detection System for Shopee.

demo.mp4

🌟 Key Features

• Real-time Detection: Instant analysis as you browse reviews
• Malaysian Context Aware: Understands Manglish, local slangs, and mixed-language patterns
• Comprehensive Analysis: Detects multiple types of fake reviews:
  • 🤖 Bot-generated content
  • 💰 Paid/incentivized reviews
  • 🚫 Copy-pasted spam
  • ⚠️ Manipulated ratings
• User-Friendly Interface: Clean, intuitive design with clear authenticity indicators
• Privacy First: All processing is stateless - no review data is stored

🚀 Quick Start

1. Download the extension:

   git clone https://github.com/Ywintersss/Frauditor.git

2. Install the Chrome extension:

   • Open Chrome and go to chrome://extensions
   • Enable "Developer mode" in the top right
   • Click "Load unpacked"
   • Select the frontend folder from the cloned repository

3. Start detecting fake reviews:

   • Visit any Shopee product page
   • Highlight any review text
   • See instant authenticity analysis!

💡 How It Works

Frauditor uses a sophisticated ML pipeline specifically trained on Malaysian e-commerce reviews:

1. Text Analysis: Advanced NLP to understand review content
2. Pattern Recognition: ML models detect suspicious patterns
3. Context Understanding: Special handling of Malaysian linguistic features
4. Risk Assessment: Multi-factor scoring system for accurate detection

🛠️ Technical Stack

• Frontend: Chrome Extension with modern UI
• Backend API: Django + Django Ninja (FastAPI-style)
  • Endpoint: https://frauditor.onrender.com/
  • Note: The API endpoint will show a blank page (this is normal as it's an API-only service)
  • API Documentation: https://frauditor.onrender.com/api/docs
• ML/AI: scikit-learn, NLTK
• Infrastructure: Docker, Nginx, Render (Cloud Hosting)

📊 Performance

• 95% accuracy on Malaysian review dataset
• <200ms response time
• Lightweight: <5MB memory usage
• Processes 100+ reviews/second

🔒 Privacy & Security

• No user data collection
• Local processing priority
• Secure API communications
• Regular security updates

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

📝 License

This project is licensed under the MIT License - see LICENSE for details.

🏆 Recognition

• Focused on solving real e-commerce challenges in Malaysia
• Developed by Team Fighting

📞 Support

For support:

• API Status: https://frauditor.onrender.com/
• Open an issue on GitHub

---

Built with ❤️ in Malaysia for Malaysian E-commerce