🧠 RevuIQ - AI-Powered Review Management System

Centralized NLP-Powered Platform for Multi-Platform Review Analysis & Response

🎯 Project Overview

RevuIQ uses Natural Language Processing to automate customer review management across Google, Yelp, TripAdvisor, and Meta platforms.

Key Features

• 📊 Sentiment Analysis - Classify reviews as Positive, Neutral, or Negative
• 😊 Emotion Detection - Identify emotional tone (anger, joy, disappointment, etc.)
• 🔍 Aspect Extraction - Detect what customers are talking about (service, food, price)
• ✍️ AI Response Generation - Create brand-consistent, empathetic replies
• 👤 Human-in-the-Loop - Approve/edit AI suggestions before posting
• 📈 Analytics Dashboard - Visualize trends and insights

🛠️ Tech Stack

Backend:

• FastAPI (Python)
• SQLite / PostgreSQL
• Hugging Face Transformers

Frontend:

• Next.js
• Tailwind CSS
• Framer Motion
• Recharts

NLP Models:

• RoBERTa (Sentiment Analysis)
• GoEmotions (Emotion Detection)
• Flan-T5 (Response Generation)
• BART/T5 (Summarization)

📦 Project Structure

RevuIQ/
├── nlp_pipeline/           # Core NLP components
│   ├── sentiment_analyzer.py
│   ├── emotion_detector.py
│   ├── aspect_extractor.py
│   ├── response_generator.py
│   └── demo.py
├── backend/                # FastAPI server
│   ├── simple_api.py      # Main API
│   ├── models.py          # Database models
│   ├── database.py        # DB configuration
│   └── google_places_integration.py
├── frontend/               # Next.js app
│   ├── app/               # Pages
│   ├── components/        # Reusable components
│   └── public/            # Static assets
├── tests/                  # Unit tests
├── requirements.txt
└── README.md

🚀 Quick Start

1. Install Dependencies

pip install -r requirements.txt
cd frontend && npm install

2. Start the Application

./start_all.sh

This will start:

• Backend API on http://localhost:8000
• Frontend on http://localhost:3000

3. Access the Dashboard

Open http://localhost:3000 in your browser and start managing reviews!

📊 NLP Pipeline Workflow

Review Input
    ↓
[Preprocessing] → Tokenization, Cleaning
    ↓
[Sentiment Analysis] → Positive/Neutral/Negative
    ↓
[Emotion Detection] → Joy, Anger, Disappointment, etc.
    ↓
[Aspect Extraction] → Service, Food, Price, Staff
    ↓
[Response Generation] → AI-generated reply
    ↓
[Human Approval] → Manager reviews & approves
    ↓
Post to Platform

🎯 Features

✅ Implemented

• Restaurant management
• Google Places API integration
• Review fetching and storage
• Sentiment & emotion analysis
• AI response generation
• Review approval workflow
• Response approval workflow
• Analytics dashboard
• Multi-page frontend
• Beautiful UI with animations

🔄 In Progress

• Advanced NLP models
• Multi-platform support (Yelp, TripAdvisor, Meta)
• Automated posting to platforms

📈 Evaluation Metrics

• Sentiment Accuracy: F1-score on labeled dataset
• Response Relevance: BLEU/ROUGE scores
• Approval Rate: % of AI replies accepted without edits
• Response Time: Average time saved vs manual handling

🤝 Contributing

This is an educational NLP project demonstrating:

• End-to-end ML pipeline design
• Transformer model integration
• Ethical AI with human oversight
• Real-world business application

📝 License

MIT License - Educational Project

🎓 Learning Outcomes

• NLP pipeline architecture
• Transformer model fine-tuning
• API design and integration
• Human-in-the-loop AI systems
• Data visualization and UX
• Full-stack development

---

Built with ❤️ for demonstrating practical NLP applications

Scripts

• ./start_all.sh - Start both backend and frontend
• ./stop_all.sh - Stop all services
• ./check_status.sh - Check if services are running

Notes

• Google Places API only returns 5 reviews per restaurant
• Reviews are analyzed using NLP
• Database is SQLite stored in backend/revuiq.db