AspectBasedSentimentAnalysis

This project focuses on performing Aspect-Based Sentiment Analysis (ABSA) on product reviews using deep learning models such as BiLSTM with Attention and Explainable AI (XAI) techniques like SHAP for interpretation. It provides fine-grained sentiment analysis for specific aspects within a review.

🎯 Goal

To accurately determine the sentiment (positive, negative, neutral) for each aspect mentioned in a customer review using an interpretable deep learning pipeline.

🔧 Technologies Used

• Python
• PyTorch
• TensorFlow (for comparison)
• NLTK / SpaCy (text preprocessing)
• BiLSTM with Attention
• SHAP (SHapley Additive exPlanations)
• Streamlit (optional: UI for demo)

📌 Features

• Fine-grained sentiment classification per aspect (e.g., "battery life", "camera", etc.)
• Attention mechanism for better context understanding
• Explainable AI using SHAP for aspect-wise transparency