Sentiment Analysis on Movie and Product Reviews Using Supervised Learning and Ensemble Learning
Year: 2025
Type: article
Abstract: This study explores sentiment analysis on IMDB (Internet Movie Database) 50K dataset and Spotify user reviews using a combination of Linear SVC (Support Vector Classifier), Ensemble Learning (Adaptive Boosting), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), enhanced by TF-IDF (Term Frequency-Inverse Document Frequency) feature extraction and hyperparameter tuning (GridSearchCV).Experimental results show that GRU with TF-IDF achieves the highest accuracy 97%, outperforming LSTM and Linear SVC with TF-IDF, which also deliver strong performance.While Adaptive Boosting (AdaBoost) with TF-IDF provides moderate accuracy, Linear SVC with ANOVA (Analysis of Variance) performs the lowest, highlighting the effectiveness of TF-IDF over ANOVA for sentiment classification.The findings confirm the superiority of deep learning models (GRU and LSTM) in handling sequential text data, while Linear SVC remains a strong, efficient alternative.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4409406876