<![CDATA[The Free Nutritious Meal Program, implemented as a national priority policy, has generated a wide array of public responses on social media, reflecting society's perception of the initiative. However, accurately classifying sentiment from short-text content such as tweets remains a challenging task due to limited context, informal language, and high lexical variability. Traditional machine learning models often fall short in capturing the complex semantic and structural information embedded in such data. To address this issue, this study proposes a sentiment classification model based on the Graph Convolutional Network (GCN) architecture, utilizing TF-IDF embeddings as the text representation method. A dataset comprising 5,979 labeled tweets was collected from the social media platform X. A document graph was constructed by computing cosine similarity between TF-IDF vector representations of the tweets, and this graph served as input to the GCN model. The proposed method was evaluated using accuracy, precision, recall, and F1-score. Experimental results show that the TF-IDF-based GCN model achieved an accuracy of 82.71% and an F1-score of 82.65%, outperforming conventional classifiers such as Logistic Regression, Support Vector Machine, and Naive Bayes. These findings demonstrate the effectiveness of integrating semantic-rich embeddings and graph-based learning in classifying public sentiment toward policy-related topics on social media.]]>
Copyrights © 2026
