<![CDATA[This study analyzes public sentiment toward the Free Nutritious Meal Program (MBG) using the Multinomial Naive Bayes algorithm on data from X (Twitter) and TikTok. A total of 5,173 entries were collected through web scraping and processed with cleaning, normalization, tokenization, stopword removal, and stemming. To address class imbalance, SMOTE was applied, and evaluation employed accuracy, precision, recall, F1-score, and AUC-ROC. Results show that without SMOTE, the model tended to be biased toward the majority class, especially on TikTok, while after SMOTE recall increased significantly and a better balance between precision and recall was achieved. On Twitter, performance was more stable with a moderate class distribution, and SMOTE further improved sensitivity to positive sentiment. Word cloud analysis revealed differences across platforms: TikTok leaned more toward negative sentiment with dominant words such as “racun,” “korupsi,” and “dapur,” while Twitter showed a stronger balance with positive terms like “gizi,” “gratis,” and “program.” These findings highlight the importance of cross-platform analysis to comprehensively understand public perceptions.]]>
Copyrights © 2026
