<![CDATA[The development of mobile application technology has driven increased user interaction through digital reviews on the Google Play Store platform. The review contains opinions that reflect the user's level of satisfaction, experience, and complaints about the app. However, the large number of reviews and variations in language expressions make manual analysis inefficient and potentially subjective. The main problem in this study is how to determine the most effective sentiment classification model to accurately identify users emotional tendencies. This study aims to compare the performance of the Naive Bayes method as a conventional machine learning model with Long Short Term Memory (LSTM) as a deep learning model based on word order in analyzing the sentiment of user reviews of Google Play Store applications. The dataset used comes from Google Play Store Reviews and goes through a pre-process process that includes text cleanup, tokenization, stopword removal, and sentiment labeling based on rating scales. The Naive Bayes model is trained using the TF-IDF representation, while the LSTM model uses an embedding sequence with standardized input padding. Evaluation uses accuracy metrics and F1-score with a ratio of 80 : 20 to train and test data distribution. The test results showed that the Naïve Bayes model achieved an accuracy of 65.78% with an F1 score of 0.5589, while the LSTM only achieved an accuracy of 45.26% with an F1-score of 0.2077. Thus, Naive Bayes was established as the best model in this study]]>
Copyrights © 2026
