Toward Intelligent Evaluation of Digital Public Services: Evidence from Indonesia’s SIM Online Platform

This investigation examines more than 65,000 user evaluations from the Google Play Store concerning sentiment toward the Digital Korlantas POLRI application. Reviews were systematically processed and categorized through Naïve Bayes, Support Vector Machine, and Random Forest classifiers, thereby highlighting the perceived strengths and weaknesses of the platform. Among the tested models, Naïve Bayes outperformed the others, yielding the greatest accuracy, precision, and recall. The analysis therefore provides quantifiable evidence of the application’s functional efficacy, the nuances of user experience, and the quality-of-service delivery. Such findings constitute actionable feedback for iterative enhancements aimed at optimizing application performance and augmenting user satisfaction. Future investigations may expand this foundational work by broadening the dataset to encompass diverse feedback channels and employing aspect-based sentiment analysis to isolate and scrutinize particular features and localized user concerns.