<![CDATA[Public sentiment analysis of the alleged corruption case of oil imports by PT Pertamina was carried out using a machine learning approach using the Extreme Gradient Boosting (XGBoost) algorithm. Data was obtained from user comments on Pertamina's official accounts on social media platforms X, Instagram, and TikTok during the period from February 1, 2025 to March 31, 2025. Comments collected through the scraping process were then processed through text preprocessing stages such as normalization, tokenization, filtering, and stemming. Feature representation was carried out using the Term Frequency-Inverse Document Frequency (TF-IDF) method to convert text data into numeric form. Each comment was manually labeled with sentiment into three categories, namely negative (0), neutral (1), and positive (2). The XGBoost model was trained with TF-IDF extracted data and evaluated using metrics such as accuracy, precision, recall, and f1-score. The evaluation results showed that the model was able to classify sentiment with good performance, with an accuracy value of 78% and an F1-score of 78%. The application of this method shows the effectiveness of the machine learning approach in understanding public perception of strategic issues in the national energy sector. The findings show that only 8% of positive sentiment on platform X, Instagram and TikTok indicates a crisis of public trust that needs to be responded to by Pertamina.]]>
Copyrights © 2025
