README.md

---
title: Advanced Sentiment Analyzer
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
---

# 🚀 Advanced Sentiment Analyzer

**Multi-Model AI System for Superior Sentiment Analysis**

This space demonstrates an advanced sentiment analysis system that uses multiple AI models working together to provide more accurate predictions than any single model alone.

## 🤖 How It Works

The system employs up to 4 different transformer models:
- **YelpReviewsAnalyzer**: Custom fine-tuned model (78.5% accuracy)
- **DistilBERT**: General-purpose sentiment analysis
- **Twitter-RoBERTa**: Optimized for social media text
- **FinBERT**: Specialized for financial sentiment

These models work together using a **consensus algorithm** that:
1. Runs all models in parallel
2. Collects individual predictions
3. Builds consensus through weighted voting
4. Provides agreement scores for reliability assessment

## 🎯 Features

- **Multi-Model Consensus**: Higher accuracy through model ensemble
- **Real-time Analysis**: Fast sentiment prediction
- **Confidence Scoring**: Know how certain the prediction is
- **Agreement Assessment**: Understand model consensus level
- **Production Ready**: Built for real-world applications

## 📊 Performance

- **Individual Model Accuracy**: 72-78%
- **Consensus Accuracy**: ~85%+ through ensemble voting
- **Processing Speed**: < 2 seconds for multi-model analysis

## 🔗 Links

- **GitHub Repository**: [Sentiment-Analyzer](https://github.com/fitsblb/Sentiment-Analyzer)
- **Original Model**: [YelpReviewsAnalyzer](https://huggingface.co/fitsblb/YelpReviewsAnalyzer)
- **Paper/Research**: Complete methodology in GitHub repo

---

*Built with ❤️ using Hugging Face Transformers, PyTorch, and Gradio*