The implementation of text mining to improve google classroom performance based on user review

The internet is developing rapidly, especially in education (e-learning). Google Classroom is an e-learning platform that has been widely used for online learning. Google Classroom gets an average rating of 3.5 out of 5 on Google Play, so it is necessary to research to increase the application's rating based on user reviews with text mining. The data used in this study was 77,454 user reviews from Google Play. Data processing uses the Textblob and Naïve Bayes Classifier methods to determine user sentiment. The sentiment analysis results with a 70/30 split data yield an accuracy of 92.39%. Data with negative sentiments is then processed with the Association Rules method to find words used as problem keywords, which include keywords' assignment', 'upload,' 'submit,' 'file,' 'class,' 'notification,' and 'dark.' The word is then analyzed using a fishbone diagram to find the root cause. The root of the problems includes problems uploading files that are integrated with Google Drive, there is no setting to change the display to 'dark mode,' The user has a good internet connection, but the file upload process is slow, and so on. The recommendations for improvements are synchronizing file integration with Google Drive, changing the display to dark mode according to user preferences, and updating the application server so that the file upload process can be done more quickly.