Movie Reviews Sentiment Analysis
This project applies sentiment analysis techniques to classify IMDb movie reviews as positive, negative, or neutral. It combines the rule-based VADER sentiment analyzer for quick polarity scoring with the transformer-based cardiffnlp/twitter-roberta-base-sentiment model for deeper contextual understanding. The workflow includes text preprocessing, model inference, and visualizations to compare the performance and insights of both traditional and modern NLP approaches.
Year
2025
Service
ML Model
Category
NLP
Tool
Hugging Face Transformers
Description:
This project focuses on analyzing the sentiment of IMDb movie reviews using both traditional and transformer-based Natural Language Processing (NLP) approaches. The workflow begins with preprocessing the text by lowercasing, removing special characters, and tokenizing. For quick rule-based sentiment scoring, the VADER sentiment analyzer is applied to capture polarity based on lexical cues.
To incorporate deeper contextual understanding, the cardiffnlp/twitter-roberta-base-sentiment transformer model from Hugging Face is used for advanced sentiment classification. Both approaches are evaluated and compared to highlight the trade-offs between lexicon-based and deep learning methods. The analysis also includes visualizing sentiment distributions and results using Matplotlib and Seaborn for clearer interpretation.
The project demonstrates a hybrid sentiment analysis pipeline that integrates classical NLP methods with modern transformer architectures, showcasing their respective strengths when applied to large-scale text datasets.
Key Highlights:
Problem: Identify the sentiment (positive, negative, neutral) of movie reviews.
Approach: VADER (rule-based) + RoBERTa Transformer (deep learning).
Dataset: IMDb Movie Reviews Dataset from Kaggle.
Deployment: Local execution in Python with visualizations for sentiment insights.
Tools:
Python
Pandas
NLTK
VADER Sentiment Analyzer
Hugging Face Transformers (
cardiffnlp/twitter-roberta-base-sentiment)Torch
Scipy
Matplotlib
Seaborn
Hugging Face Transformers
