<![CDATA[This study aims to classify reviews of the SIREKAP 2024 application automatically using the DistilBERT feature extraction method and the Support Vector Machine (SVM) algorithm. The data used includes 8,538 user reviews from the Google Play Store with five Rating categories as the target variable. After undergoing 10-Fold cross-validation, the average F1-Score obtained was 36.62%, with the highest performance reaching 37.16%. The analysis indicates that data imbalance is the main obstacle in improving the model's accuracy, particularly in the minority class. The study concludes that the combination of DistilBERT and SVM yields suboptimal results and requires further optimization. Recommendations are provided to improve model accuracy and enhance the quality of the application based on user reviews.]]>
Copyrights © 2025
