Analisis Sentimen Layanan Publik KRL Commuter Line Menggunakan Algoritma Support Vector Machine (Studi Kasus Media Sosial X)

Social media platform X has become a means for the public to express opinions about the quality of KRL Commuter Line services. Therefore, sentiment analysis is needed to systematically assess users’ perceptions. This study aims to analyze user sentiment toward KRL Commuter Line services using the Support Vector Machine (SVM) algorithm, with social media platform X as the case study. The research data were collected through web scraping, resulting in 6,267 tweets, which were processed through text preprocessing stages, including filtering, case folding, cleaning, and normalization. Data labeling was conducted automatically using a hybrid approach that combined rule-based keyword matching and Zero-Shot Classification with the mDeBERTa-v3 model. Subsequently, text features were extracted using contextual embedding from mDeBERTa-v3 and classified using SVM with a One-vs-Rest (OvR) scheme, while parameter optimization was performed using GridSearchCV. Model evaluation employed accuracy, precision, recall, and F1-score metrics. The results showed that the SVM model achieved its best performance in classifying user sentiment regarding cleanliness, with an accuracy of 82% and an F1-Macro of 81.55%, followed by punctuality, with an accuracy of 69% and an F1-Macro of 68.94%, and comfort, with an accuracy of 73% and an F1-Macro of 53.54%. These findings indicate that SVM is effective in analyzing sentiment toward KRL Commuter Line services on social media platform X, although its performance is still influenced by the distribution of data across sentiment classes.