<![CDATA[The rapid growth of social media has transformed public communication patterns and positioned platform X as a digital space where citizens actively express their views on government policies, including the Free Nutritious Meal Program (Program Makan Bergizi Gratis/MBG). As a strategic national initiative aimed at improving students’ nutritional quality, the implementation of the MBG Program has generated diverse public responses that require systematic analysis. This study aims to identify public sentiment tendencies toward the MBG Program and to compare the classification performance of Support Vector Machine (SVM) and Naïve Bayes algorithms in sentiment analysis based on social media text. The research data consist of Indonesian-language tweets collected through a web scraping process using keywords related to the MBG Program. The collected data were processed through several text preprocessing stages to reduce noise and enhance data quality. Sentiment labeling was conducted automatically using a lexicon-based approach, classifying tweets into positive, neutral, and negative categories. Feature representation was performed using the Term Frequency–Inverse Document Frequency (TF-IDF) method, and the dataset was divided into training and testing sets with an 80:20 ratio. Sentiment classification was then carried out using SVM and Naïve Bayes algorithms, with model performance evaluated based on accuracy metrics. The experimental results show that the SVM algorithm achieved an accuracy of 87.57%, outperforming the Naïve Bayes algorithm, which obtained an accuracy of 68.08%. These findings indicate that SVM is more effective in handling high-dimensional and unstructured social media text data]]>
