Sentiment Classification of User Reviews on a Digital Public Service Application

Abstract

Advances in information technology have encouraged local governments to improve the quality of public services through the use of digital innovations. One such innovation is the Public Service Mall (Mall Pelayanan Publik/MPP), designed to facilitate the public in accessing various government services in an integrated manner. The Sumedang Regency Government developed the MPP Sumedang Application as a digital platform that consolidates multiple regional public services. However, user responses to this application vary considerably, as reflected in reviews posted on the Google Play Store. This study aims to classify user sentiments toward the MPP Sumedang Application based on these reviews. The research methods include data collection, text preprocessing, feature weighting using Term Frequency–Inverse Document Frequency (TF–IDF), and sentiment classification using the Naive Bayes algorithm into three categories: positive, negative, and neutral. The test results show a model accuracy of 80%, with a precision of 0.74 and an F1-score of 0.86. From a total of 500 reviews, 429 reviews (85.8%) were classified as negative, 51 (10.2%) as positive, and 20 (4%) as neutral. The dominance of negative sentiment indicates that issues related to application performance and ease of use are still present. These findings illustrate public perceptions of the MPP Sumedang Application and serve as input for the local government in improving the quality of digital public services.