| | ---
|
| | title: ABSA
|
| | app_file: app_spaces.py
|
| | sdk: gradio
|
| | sdk_version: 5.9.1
|
| | ---
|
| |
|
| | # 🍽️ Restaurant Review Analyzer
|
| |
|
| | A Gradio-powered web interface for analyzing restaurant reviews using **Aspect-Based Sentiment Analysis (ABSA)**. This application identifies specific aspects (like food, service, atmosphere) mentioned in reviews and determines the sentiment associated with each aspect.
|
| |
|
| | ## 🎯 How It Works
|
| |
|
| | The application uses two fine-tuned DistilBERT models:
|
| |
|
| | 1. **Aspect Extraction**: Identifies aspects mentioned in reviews (e.g., "food", "service", "atmosphere")
|
| | 2. **Sentiment Classification**: Determines sentiment (positive/negative) for each aspect
|
| |
|
| | ## 🚀 Try It Out!
|
| |
|
| | Simply enter a restaurant review in the text box and click "Analyze Sentiment" to see:
|
| | - **Identified Aspects**: What specific elements are mentioned
|
| | - **Sentiment Analysis**: Whether each aspect is viewed positively or negatively
|
| | - **Confidence Scores**: How certain the model is about each prediction
|
| |
|
| | ## 📊 Example
|
| |
|
| | **Input**: "The services here is wonderful, but I hate the food. However, I still love the atmosphere here."
|
| |
|
| | **Output**:
|
| | - **service** → POSITIVE (0.952)
|
| | - **food** → NEGATIVE (0.887)
|
| | - **atmosphere** → POSITIVE (0.934)
|
| |
|
| | ## 🔧 Models
|
| |
|
| | - **Aspect Extraction**: [sdf299/abte-restaurants-distilbert-base-uncased](https://huggingface.co/sdf299/abte-restaurants-distilbert-base-uncased)
|
| | - **Sentiment Classification**: [sdf299/absa-restaurants-distilbert-base-uncased](https://huggingface.co/sdf299/absa-restaurants-distilbert-base-uncased)
|
| |
|
| | ## 💡 Use Cases
|
| |
|
| | Perfect for:
|
| | - Restaurant owners analyzing customer feedback
|
| | - Review aggregation platforms
|
| | - Market research on dining experiences
|
| | - Academic research in sentiment analysis
|
| | - Understanding customer opinions at scale
|
| |
|
| | ---
|
| |
|
| | *Built with 🤗 Transformers and Gradio* |
