<![CDATA[The Free Nutritious Meal Program (MBG), which represents a priority program of President Prabowo Subianto, has garnered widespread attention from Indonesian society. This program has received sympathy from various groups, including students and informal workers, and has been extensively discussed through social media, particularly on platform X. This research aims to analyze public response in the form of positive and negative sentiment toward the MBG Program based on data from platform X. A total of 1,378 tweets were collected using crawling methods, followed by preprocessing, sentiment labeling using the InSet lexicon dictionary, and feature extraction using three techniques: Term Presence, Bag of Words (BoW), and Term Frequency-Inverse Document Frequency (TF-IDF). Subsequently, sentiment classification was performed using the Support Vector Machine (SVM) algorithm for each feature extraction technique. Classification results demonstrate that the TF-IDF technique achieved the highest accuracy of 77.5%, compared to Term Presence (76.2%) and BoW (75.3%). Validation using K-Fold Cross Validation with five iterations was conducted with imbalanced data handling through the Synthetic Minority Over-sampling Technique (SMOTE) method. In this validation, TF-IDF consistently demonstrated superior performance with an average accuracy of 75.54%, precision of 74.31%, recall of 73.86%, and f1-score of 73.98%. Despite a slight decrease in accuracy following data synthesis, the TF-IDF technique proved to be stable and effective in handling data variation. The superiority of the TF-IDF feature extraction technique is suitable for combination with the SVM algorithm.]]>
