Sentiment Analysis of TikTok User Reviews on the Google Play Store Using the Naïve Bayes Classifier Method

The high volume of reviews provided by TikTok users on the Google Play Store platform has resulted in a large amount of unstructured opinion data. This study analyzed user sentiment to provide developers with insights into user satisfaction levels and common application complaints. The method implemented in this research was the Multinomial Naïve Bayes classifier. The dataset consisted of 860 application reviews collected over a two-year period, from 2023 to 2024. All data were processed through a comprehensive series of text mining stages using the Python programming language, which included text preprocessing (case folding, cleaning, tokenizing, filtering, and stemming) and feature weighting via Term Frequency-Inverse Document Frequency (TF-IDF). The test results demonstrated that the proposed model successfully classified sentiments into positive, negative, and neutral categories with an overall accuracy level of 79%. Specifically, the model achieved its optimal performance in the positive sentiment category with an f1-score of 0.86. These findings indicate that the Naïve Bayes method has reliable capability in processing social media reviews for continuous application quality improvement.