Sentiment Analysis of Social Media for Indonesian m-Health PeduliLindungi Mobile-Apps(PLMA) with Lexicon-Based and Support Vector Machine Approach

Indonesia launched the e-health app PeduliLindungi Mobile-Apps (PLMA) in April 2020 to monitor COVID-19, gaining public confidence with over ten million downloads and rapid growth. However, PLMA faces challenges like application errors, leaks, and inaccurate user data, with 52% of Indonesian users reporting errors. Despite being Indonesia's most downloaded medical app, its rating remains low. User sentiment research is crucial for improving app quality and functionality. Study voids include insufficient research and local context. This research examines TikTok, YouTube, and Twitter user perceptions of the Indonesian mHealth app PLMA using Lexicon-based and SVM analysis, reducing bias and errors to enhance the study. The result shows that PLMA perception on YouTube, TikTok, and Twitter uses Lexicon-based and Support Vector Machine sentiment analysis, with 86% accuracy on TikTok and 64% on Twitter. However, Lexicon-based and Naive Bayes algorithms struggle with Twitter data classification. TikTok sentiment classification is 8% negative, 69% neutral, and 23% positive, while YouTube and Twitter have 24% unfavorable, 14% neutral, and 62% positive opinions, respectively. This research aids developers, stakeholders, and policymakers in understanding app strengths and weaknesses, improving user experience, and developing advanced sentiment analysis techniques, thereby enhancing knowledge and health sector decision-making.