<![CDATA[Social media, particularly TikTok, has become a strategic platform for politicians to build personal branding and shape public opinion. This study aims to analyze public sentiment toward Dedi Mulyadi’s personal branding on TikTok and to compare sentiment analysis results obtained through manual labeling and data mining methods using the Naïve Bayes algorithm. This research employs a quantitative descriptive–comparative approach. The dataset consists of 800 manually labeled comments and 3,226 comments collected through web scraping and automatically classified. The results show that sentiment analysis using the manual method produces positive sentiment as the dominant category, accounting for 55.4%, followed by neutral sentiment at 30.6% and negative sentiment at 14.0%. In contrast, the data mining–based sentiment analysis indicates neutral sentiment as the most dominant category at 45.66%, followed by negative sentiment at 28.51% and positive sentiment at 25.83%. These differences are influenced by data volume, labeling techniques, and the limitations of algorithms in interpreting informal language and implicit expressions. This study concludes that both manual and data mining approaches have distinct strengths and limitations; therefore, their combined use can provide a more comprehensive understanding of public sentiment toward political personal branding on social media.]]>
Copyrights © 2026
