Comparative Study of ML Models for Movie Sentiment Classification
A new arXiv paper (2605.22003) compares machine learning models for sentiment classification of movie reviews using the IMDb dataset. The study evaluates Naive Bayes, Logistic Regression, SVM, LightGBM, LSTM, and transformer-based models RoBERTa and DistilBERT. The authors note that ML models struggle with context and metaphysical sentiment due to reliance on statistical word representations. The paper employs NLP methodologies for preprocessing and assessment.
Key facts
- arXiv paper 2605.22003 compares ML models for sentiment classification
- Uses IMDb dataset for movie review sentiment analysis
- Models evaluated: Naive Bayes, Logistic Regression, SVM, LightGBM, LSTM, RoBERTa, DistilBERT
- ML models struggle with context and metaphysical sentiment
- NLP methodologies used for preprocessing and assessment
- Objective: classify movie reviews as positive or negative
- Sentiment analysis also called opinion mining
- Paper is a cross-type announcement on arXiv
Entities
Institutions
- arXiv