<![CDATA[Abstract: The rapid growth of short-video social media platforms such as TikTok has significantly increased the volume of user reviews that reflect public perceptions of application quality. These reviews constitute electronic word of mouth (e-WOM), which influences brand image, user trust, and adoption decisions. This study aims to analyze user sentiment toward the TikTok application using a text mining approach based on the Naïve Bayes algorithm and Term Frequency–Inverse Document Frequency (TF-IDF) weighting. The dataset consists of user reviews collected from the Google Play Store and categorized into sentiment classes. The data were processed through several preprocessing stages, including text cleaning, tokenization, stopword removal, normalization, and stemming, before feature extraction and classification. The experimental results indicate that the proposed model achieved an accuracy of 67.35% in classifying sentiment. However, analysis of the confusion matrix and prediction distribution reveals a bias toward the majority class due to dataset imbalance. The Naïve Bayes classifier combined with TF-IDF representation demonstrates satisfactory performance in identifying dominant extreme sentiments (positive and negative), yet its effectiveness decreases in multi-class classification scenarios with uneven data distribution.]]>
Copyrights © 2026
