<![CDATA[Diverse public opinions on social and economic issues related to labor migration are often expressed on the social media platform X (Twitter). This research aims to classify public sentiment toward this phenomenon by analyzing tweets containing the hashtag "#KaburAjaDulu". Sentiment classification is performed by comparing two Support Vector Machine (SVM) approaches that utilize indoBERT embeddings, a language model designed to capture the nuances of the Indonesian language. Both SVM models are trained using web crawling data from the X platform, with the main difference lying in the application of hyperparameter tuning on one of the models. The data collected through web crawling from the X platform then undergoes a pre-processing stage that includes text normalization and stopword removal. The results show that the SVM model optimized through hyperparameter tuning achieved an accuracy of 90.5%, higher than the SVM model without tuning which achieved only 77.7%. This finding underscores the importance of hyperparameter tuning in improving the performance of sentiment classification models, especially when utilizing rich feature representations such as indoBERT embeddings to understand deeper language context.]]>
Copyrights © 2025
