Analisis Sentimen Aplikasi Samsat Digital di Play Store Menggunakan Support Vector Machine

Abstract

Technology plays a crucial role in the development of public services in today's digital era. The National Digital Samsat application (SIGNAL) is a tangible example of the application of information technology aimed at facilitating the payment process for motor vehicle taxes. By utilizing digital systems, the public can conduct transactions online without the need to physically visit the Samsat office. This study aims to analyze user sentiment regarding the SIGNAL application on the Play Store using the Support Vector Machine (SVM) method. The evaluation was conducted by comparing the accuracy of four SVM kernels-linear, RBF, sigmoid, and polynomial, across three scenarios. The best results were obtained in scenario 1, with a training data ratio of 90% and testing data of 10%, yielding an accuracy of 92.2%, precision of 97.21%, recall of 97.2%, and an F1-score of 97.19%. The analysis also indicates that the majority of user reviews are positive, with 531 positive reviews and 219 negative reviews. The words that frequently appear in positive reviews include "application," "easy," "tax," and "help." These findings suggest that the SIGNAL application has been well received by users and can serve as a foundation for developers to continually improve the service.