Review Analysis of SatuSehat Application Using Support Vector Machine and Latent Dirichlet Allocation Modeling

Abstract

SatuSehat is a contact tracing application that replaces the PeduliLindungi application initiated by the Government of Indonesia with the aim of tracking the Covid -19 Virus. The success of the application can be known by analyzing sentiment reviews. In addition to the high number of reviews, there are also other things that need to be highlighted, namely the pattern of reviews that are not in accordance with refined spelling and diverse topics, so that identifying a topic from a collection of reviews is very difficult and takes a lot of time if done manually by humans. This research describes sentiment analysis and topic modeling on SatuSehat app user reviews. By applying Support Vector Machine (SVM) method for sentiment analysis and Latent Dirichlet Allocation (LDA) for topic modeling, this study reveals the views and trends expressed by users. The analyzed review data from Google Play Store includes 171,428 positive reviews and 131,246 negative reviews. The sentiment analysis results indicated the dominance of positive responses. LDA modeling resulted in 8 identified topics, from health concerns to app appreciation. However, negative topics included vaccination challenges, access issues, and app functionality. This research provides insight into users' perceptions of the SatuSehat app, providing a basis for further development and improvement of the app.