<![CDATA[The rapid growth of social media usage has led to a continuous increase in the volume of user reviews, necessitating automated analysis based on machine learning techniques. This study focuses on the development of a sentiment classification model for Pinterest application reviews on the Google Play Store by evaluating three algorithms: Support Vector Machine (SVM), Naïve Bayes, and K-Nearest Neighbors (KNN), combined with Synthetic Minority Oversampling Technique (SMOTE) and Particle Swarm Optimization (PSO). A total of 10,000 reviews were collected through web scraping and processed through preprocessing stages, lexicon-based labeling using InSet, TF-IDF feature extraction, and an 80:20 data split. SMOTE was first applied to balance the class distribution, followed by PSO for parameter optimization of each classification algorithm. The experimental results indicate that SVM achieved the best performance, attaining 95% accuracy with a more balanced F1-score after the application of SMOTE and PSO, while Naïve Bayes and KNN remained sensitive to class imbalance. As the final output, this study developed a Streamlit-based prediction dashboard to display sentiment results in real time, thereby supporting practical and efficient analysis of user perceptions. These findings confirm the effectiveness of combining SVM, SMOTE, and PSO as an optimal approach for sentiment classification on imbalanced review data. ]]>
Copyrights © 2026
