<![CDATA[The rapid growth of internet use in Indonesia has contributed to the rise of cyberbullying on TikTok, increasing the importance of automated sentiment analysis for digital safety. This study compares the performance of Support Vector Machine, K-Nearest Neighbors, and Naive Bayes in classifying sentiments in TikTok comments related to cyberbullying. The dataset was collected via web scraping and processed through several preprocessing stages, yielding 7,900 unique comments. Sentiment labeling used a lexicon-based approach, and the data were split into training and testing sets with an 80:20 ratio. Results show that 34.18% of comments were negative, indicating a notable level of harmful content. Among the three models, Support Vector Machine performed best with an accuracy of 91.5%, followed by Naive Bayes at 82.8% and K-Nearest Neighbors at 80.8%. These findings suggest Support Vector Machine is the most effective method for sentiment classification in this context and offer a useful reference for developing more accurate content moderation systems on social media.]]>
Copyrights © 2026
