<![CDATA[The Free Nutritious Meal Program (MBG) is a strategic initiative by the Indonesian government to improve child nutrition and prevent stunting. However, its implementation has sparked diverse public opinions on social media, which are difficult to analyze manually due to the large volume of data. This study aims to identify public sentiment toward the MBG program through social media X by implementing the IndoBERT model. A total of 4,380 tweets were collected using web scraping techniques with relevant keywords between March and May 2025. The research process included preprocessing (data cleaning, stopword removal, stemming, and tokenization), semi-automatic data labeling, and data division into a 71.97% training set, 8.02% validation set, and 20.01% test set. The model used was the Indonesian RoBERTa Base Sentiment Classifier architecture, which underwent a fine-tuning process for 20 epochs. The results showed that the IndoBERT model achieved an accuracy rate of 80.11% and a weighted average F1-score of 0.8000. Negative sentiment was detected most accurately with an F1-score of 0.8301. Although effective, the model still faces challenges in handling linguistic ambiguity in neutral sentiment and the risk of overfitting. Further research is recommended to expand slang language normalization and apply stricter model regulation techniques.]]>
Copyrights © 2026
