<![CDATA[The issue of “17+8 People’s Demands” that emerged within Indonesia’s socio-political dynamics has become a major topic of public discussion on social media. This viral phenomenon has generated a large volume of unstructured textual data, predominantly written in informal Indonesian and slang, thereby requiring an analytical approach capable of comprehending linguistic context more effectively. This study aims to analyze and classify social media users’ sentiments from platform X using the Twitter API. The collected texts were cleaned from noise, labeled into three sentiment categories—positive, neutral, and negative—and processed using the IndoBERT algorithm to classify the polarity of public opinion. A total of 7,936 text data were successfully obtained through a crawling process. The prepared data underwent a series of preprocessing stages before being used to evaluate the model’s performance. Overall, the evaluation results showed an accuracy of 87%. Specifically, in the aspect of class-level classification, the model demonstrated consistent performance with 90% precision, 97% recall, and an F1-score of 94%. These findings indicate the effectiveness of the IndoBERT model in accurately identifying and classifying public opinions expressed in the Indonesian language. The main contribution of this research lies in the application of a transformer-based Indonesian language model to analyze emerging social issues within digital public discourse.]]>
Copyrights © 2026
