<![CDATA[The government’s Housing Savings Program (TAPERA) has sparked various public reactions, particularly on social media platform X (Twitter). This study aims to analyze user sentiment toward the TAPERA policy using the Convolutional Neural Network (CNN) and Bidirectional Encoder Representations from Transformers (BERT) methods. The dataset was collected using a crawling technique on X (Twitter), comprising a total of 1,790 tweets. These data were processed through preprocessing stages, including cleaning, case folding, normalization, tokenization, stopword removal, and stemming. The CNN and BERT models were then trained and tested to classify sentiments as positive or negative. The findings indicate that the BERT model outperformed CNN, achieving a higher accuracy of 86% compared to CNN’s 85%, along with superior recall, precision, and F1-score values. These results suggest that the BERT method is more effective in comprehensively understanding sentiment context.]]>
Copyrights © 2025
