Analisis Sentimen Kebijakan Makan Bergizi Gratis pada Media Sosial X Menggunakan IndoBERT

Abstract

Abstract – The Free Nutritious Meals policy introduced by the government has garnered mixed reactions from the public. While some members of the public view it as a positive step toward improving children’s nutritional well-being, others have voiced concerns regarding its fiscal impact and implementation challenges on the ground. This debate is actively taking place on the social media platform X, making it a rich source of large-scale data for analysis. This study aims to classify public sentiment toward the policy in order to comprehensively map citizens’ responses. The methodology involves collecting 17,752 text data points from the X platform, which undergo preprocessing stages including data cleaning and text normalization. Subsequently, the data was classified into positive, negative, and neutral sentiment classes using the IndoBERT (Bidirectional Encoder Representations from Transformers) deep learning model, which excels at capturing semantic context. Test results show that the IndoBERT model is capable of performing classification well, as evidenced by metrics of 82% Accuracy, 81% Precision, 80% Recall, and an F1-Score of 80%. The majority of netizen discourse tends to fall into the negative category (39.3%), dominated by concerns regarding the allocation of state funding, followed by neutral (34.7%) and positive (26.0%) sentiments. In conclusion, this IndoBERT-based sentiment analysis has proven effective in capturing public opinion, thereby providing empirical insights for the government in developing policy communication strategies.