Analisis Sentimen Pada Calon Presiden Dalam Media sosial X Menggunakan Algoritma Bidirectional Encoder Representations from Transformers (BERT)

The selection of presidential candidates in the context of democratic events has a significant impact on the political and social dynamics of a country. Social media, especially platform X, such as Twitter, Facebook, and Instagram, serves as the primary channel for presidential candidates and political parties to interact with potential voters. However, the use of social media also brings risks, particularly in the form of hate speech that can trigger societal polarization and escalate social tensions. This research focuses on sentiment analysis of presidential candidates on social media X using the Bidirectional Encoder Representations from Transformers (BERT) algorithm. With the increasing role of social media in political expression, understanding sentiment becomes crucial. The testing method involves crawling data from social media X and text preprocessing before applying the BERT model. The test results show that BERT achieves an accuracy of approximately 63%. Recommendations for future research include adjusting parameters and exploring data augmentation techniques. This study provides insights into political sentiment analysis in the digital era and underscores the potential of the BERT model as a foundation for further research in social media sentiment analysis.