<![CDATA[Analysis of user reviews can provide valuable insights for app developers in improving quality, but conventional sentiment analysis only categorizes sentiment in general terms. Aspect-based Sentiment Analysis (ABSA) is a method that can be used to extract specific opinions from various aspects of user reviews. This study compares ABSA on user reviews of Ruangguru app on Android and iOS platforms. Review data was collected from Google Play Store and Apple App Store, processed, and classified into sentiment polarity using deep learning models such as BiLSTM with Word2Vec. The analysis was conducted to find out the aspects talked about by users and the sentiment associated with each aspect. The evaluation results show that the BiLSTM model with Word2Vec features performs well on the sentiment analysis task achieving 84% accuracy. In the aspect extraction task, the model performs very well with accuracy, precision, recall, and F1 Score values of 97%. These results show that the combination of BiLSTM and Word2Vec is an effective approach in understanding user opinions and preferences from Ruangguru app review text data and has the potential to be applied in the development of automated opinion analysis systems. Price aspect extraction results are the most dominant topic discussed on both platforms, followed by features and materials. Positive sentiment towards the price aspect dominates, but there is also a significant proportion of negative sentiment, especially on the Android platform.]]>
Copyrights © 2025
