<![CDATA[This study aims to analyze user sentiment toward the Stimuler AI: English Fluency application based on reviews collected from the Google Play Store using the IndoBERT model. The increasing popularity of AI-based language learning applications has generated numerous user reviews containing opinions, experiences, and complaints. Therefore, sentiment analysis is needed to understand user perceptions and evaluate application quality efficiently. This research employed a quantitative approach using sentiment analysis techniques. User reviews were collected through web scraping using Python and the google-play-scraper library. A total of 3,640 reviews were initially collected, and after language filtering, 1,653 Indonesian-language reviews were selected as the final dataset. The preprocessing stages included case folding, cleaning, stopword removal, and slang normalization. Exploratory data analysis and topic analysis were also conducted to identify dominant sentiment patterns and user concerns. The sentiment classification process utilized the IndoBERT model with three sentiment classes: positive, neutral, and negative. The results showed that positive sentiment dominated the dataset, indicating that most users were satisfied with the application’s learning features and speaking practice capabilities. Meanwhile, the main negative issues included bugs, subscription problems, and voice detection issues. Furthermore, the IndoBERT model achieved an accuracy score of 90.57% and an F1-score of 89.01%, indicating effective performance in classifying Indonesian-language application reviews.]]>
Copyrights © 2026
