<![CDATA[The number of user reviews for the iPusnas application on the Google Play Store continues to increase, making the manual sentiment analysis process difficult. In addition, traditional classification methods still have limitations in understanding the context of the Indonesian language, especially in reviews that use informal language. This study aims to analyze the sentiment of user reviews of the iPusnas application using the IndoBERT method. The research data was obtained through web scraping using Python and resulted in 3,513 reviews from January 2021 to May 2026. The research stages include text preprocessing, sentiment labeling using the pretrained RoBERTa model, splitting the data into training and testing sets, data augmentation, classification using IndoBERT, and model evaluation using a confusion matrix, accuracy, precision, recall, and F1-score. The test results show that the model achieved an accuracy of 0.79. The best F1-score was obtained in the negative sentiment class at 0.85, while the neutral and positive classes each received a score of 0.70. The research results indicate that the IndoBERT method is capable of classifying the sentiment of iPusnas application reviews well and can serve as an evaluation tool for the National Library of the Republic of Indonesia to improve the quality of iPusnas application services.]]>
Copyrights © 2026
