<![CDATA[Objective: This study aims to analyze user sentiment toward the InDrive application on Google Play Store by employing Support Vector Machine (SVM) and Naïve Bayes algorithms, motivated by the increasing number of user reviews that are difficult to evaluate manually, thus requiring a text mining approach to efficiently classify opinions into positive and negative categories. Method: A dataset of 30,000 reviews was collected through web scraping, and the analysis involved several stages, including preprocessing (cleaning, case folding, normalization, tokenizing, stopword removal, and stemming), term weighting using TF-IDF, and classification using SVM and Naïve Bayes. Results: The results revealed that SVM outperformed Naïve Bayes with an accuracy of 78%, precision of 0.80, and recall of 0.74, whereas Naïve Bayes achieved 76% accuracy, 0.79 precision, and 0.70 recall, indicating that SVM is more effective in handling complex user review data compared to Naïve Bayes. Novelty: The novelty of this research lies in applying a comparative study of the two algorithms to InDrive application reviews, which has not been extensively explored, and is expected to provide insights for developers to better understand user perceptions and improve the quality of application services.]]>
Copyrights © 2025
