<![CDATA[This study aims to analyze public sentiment regarding TikTok’s privacy and data security policies using the Multinomial Naive Bayes algorithm. Research data were obtained from a public Kaggle dataset containing 100,000 reviews from Indonesian TikTok users on the Google Play Store. After filtering using keywords related to privacy and data security, 9,590 relevant reviews were collected. The research stages include text preprocessing, TF-IDF feature extraction, SMOTE implementation to handle class imbalance, Naive Bayes modeling along with comparison against Linear SVM and IndoBERT, and model evaluation. The results show negative sentiment dominates at 59.7%, followed by positive at 30.1% and neutral at 10.1%. The Naive Bayes model achieves 69.6% accuracy, 64.2% precision, 69.6% recall, and 63.5% F1-Score, while Linear SVM and IndoBERT achieved 74.3% and 81.2% accuracy respectively. This study proves that Naive Bayes is applicable for real-data sentiment classification of TikTok privacy reviews and provides insights for platform managers to improve privacy policy transparency.]]>
Copyrights © 2026
