Sentiment Analysis of Halodoc Application Reviews Based on Service Quality Aspects Using Bert

This research is an implementation of Aspect-Based Sentiment Anaysis (ABSA) using the Bidirectional Encoder Representations from Transformers algorithm to analyze Halodoc application reviews. The reviews used in this study were extracted using the Google Play Scraper from the Google Play Store. Halodoc is a digital healthcare platform that provides health services through a mobile application and website. The platform enables users to connect with doctors and medical professionals online for medical consultations, health examinations, and prescriptions. In the ABSA implementation, aspects were identified. The aspects used in this study are Service Quality, which is divided into responsiveness, friendliness, clarity, ease of use, and security. The results of this research indicate that responsiveness has an accuracy of 86%, ease of use has an accuracy of 87%, friendliness has an accuracy of 83%, clarity has an accuracy of 95%, and security has an accuracy of 90%. The findings demonstrate BERT (Bidirectional Encoder Representations from Transformers) ability to classify user sentiments according to specific aspects quite effectively. This research contributes to understanding user sentiments towards the Halodoc application and can serve as a reference for further studies. However, the study has limitations that can be considered for future development, such as the quantity of tested data.