<![CDATA[The launch of the Indonesian Standard Quick Response Code (QRIS), scheduled for August 17, 2025, in Japan and China, requires an evaluation of public response to support the development of cross-border digital payment policies. This research aims to analyze Indonesian public sentiment toward QRIS usage in Japan and China through TikTok comments, compare the performance of the Support Vector Machine (SVM) model against the Bidirectional Encoder Representations from Transformers (BERT) approach, and provide methodological recommendations for fintech sentiment analysis in Indonesia. The study employs a quantitative approach, utilizing 1,429 TikTok user comments collected through web scraping. The data underwent cleaning, preprocessing, and labeling using TextBlob. The data were then split using a 5-fold cross-validation scheme and implemented on an SVM model with TF-IDF representation and the SMOTE technique, as well as a BERT model with data augmentation and the application of class weights. The evaluation results show that BERT achieved an accuracy of 84% with evaluation metric values ranging from 0.84 to 0.85, while SVM achieved an accuracy of 80% with consistently stable evaluation metric values of 0.80. The research confirms that deep learning approaches based on pretrained language models are optimal for social media sentiment analysis in fintech. BERT can support policymakers in monitoring public sentiment in real-time for international digital payment service development. ]]>
Copyrights © 2026
