<![CDATA[Tantan, as a popular dating application in Indonesia, has garnered various user reviews reflecting their experiences. This study aims to analyze user sentiment for the Tantan application by comparing the performance of Naive Bayes and Support Vector Machine (SVM) algorithms in sentiment classification. User reviews were collected from Google Play Store using web scraping techniques and processed through data cleaning, tokenization, and TF-IDF feature extraction. The dataset comprises 1,195 reviews, with 74.6% positive and 25.4% negative sentiments. The Naive Bayes model achieved an accuracy of 85.36%, excelling in detecting positive reviews (precision 86%, recall 97%). However, its performance on negative reviews was suboptimal, with a recall of only 44%. Conversely, the SVM model with a sigmoid kernel demonstrated superior overall performance, achieving an accuracy of 87.03%. It handled negative reviews better, with a recall of 67% and an F1-score of 69%, while maintaining excellent results for positive reviews (precision 91%, F1-score 92%). The results indicate that although both algorithms have their strengths, SVM with a sigmoid kernel is recommended for this dataset due to its balanced and stable performance. This model provides valuable insights for feature development and quality improvement strategies for the application.]]>
Copyrights © 2025
