📊 Sentiment Analysis using VADER and roBERTa

A comparative NLP project that analyzes sentiment in textual reviews using both lexicon-based (VADER) and transformer-based (roBERTa) approaches.

📋 Overview

This project compares two sentiment analysis techniques on a large corpus of text reviews:

Approach	Model	Type
VADER	Valence Aware Dictionary and sEntiment Reasoner	Lexicon-based, rule-based
roBERTa	Robustly optimized BERT pretraining approach	Transformer-based, deep learning

Both models produce polarity scores (positive, negative, neutral) for comparative analysis.

📊 Dataset

• Size: ~500,000 textual reviews
• Task: Sentiment classification and polarity scoring

⚠️ The dataset is too large to host on GitHub (even when compressed). Please obtain the review dataset from your course materials or the original source.

🏗️ Project Structure

├── Sentiment.ipynb    # Main notebook with implementation and analysis
└── README.md

📈 Contents

The notebook includes:

• Data preprocessing and exploratory analysis
• Data distribution visualizations
• VADER sentiment scoring and polarity outputs
• roBERTa sentiment scoring and polarity outputs
• Model comparison and polarity score analysis
• Limitations of each approach

🚀 Getting Started

Prerequisites

pip install pandas numpy transformers torch vaderSentiment

Usage

1. Add your review dataset to the project directory
2. Open Sentiment.ipynb in Jupyter
3. Update the data path in the notebook
4. Run all cells to preprocess, analyze, and compare sentiment scores

📝 Key Findings

• Comparative performance of lexicon-based vs. transformer-based sentiment analysis
• Polarity score outputs from both VADER and roBERTa
• Discussion of strengths and limitations of each approach

👤 Author

Ivaylo Papazov

📄 License

This project is available for educational purposes.

---

⭐ If you find this project useful, please consider giving it a star!