<![CDATA[The increasing use of social media as a public space has encouraged the emergence of various opinions on government policies, including the Free Nutritious Food Program which is widely discussed on Platform X (Twitter). However, unstructured text data and diverse user perspectives pose challenges in accurately identifying sentiment. This study aims to analyze public sentiment using the Support Vector Machine (SVM) method with Term Frequency – Inverse Document Frequency (TF-IDF) weighting. Data were collected through web scraping from August to November 2025 totaling 4,002 tweets, which were then processed through labeling and preprocessing to obtain 3,129 data. Testing was carried out with three classification scenarios, namely three classes, two classes, and positive and non-positive. The results show that the highest accuracy obtained in the positive vs. non-positive scenario is 90.57%, followed by two classes at 90.34%, and three classes at 80.67%. These findings indicate that simplifying the number of classes can improve model performance. The SVM method with TF-IDF has proven effective in sentiment analysis on social media data.]]>
Copyrights © 2026
