<![CDATA[The policy of West Java Governor Dedi Mulyadi (KDM) to send troubled youths to military barracks as a form of character development to address juvenile delinquency triggered diverse public opinions in YouTube comment sections, which were difficult to analyze manually due to their large volume and linguistic variation. This study aims to analyze the sentiment of YouTube comments toward the policy using the Naïve Bayes Classifier algorithm with Term Frequency-Inverse Document Frequency (TF-IDF) weighting, and to evaluate model performance based on accuracy, precision, recall, and F1-score metrics. The research data were obtained from a public Kaggle dataset comprising 7,875 comments, which after preprocessing resulted in 7,407 valid data with a proportion of 56.6% negative and 43.4% positive comments. The data were split using an 80:20 ratio with stratified sampling. Test results show that the Naïve Bayes model achieved an accuracy of 73.95%, precision of 84.64%, recall of 48.83%, and F1-score of 61.93%, with the negative class performing better (F1-score 80%) than the positive class (F1-score 62%) due to class imbalance in the dataset. Word cloud analysis revealed that negative sentiment was largely not directed at the KDM policy itself, but rather at criticism toward the Indonesian Child Protection Commission (KPAI). This study provides an objective overview of public perception that can serve as a basis for evaluating juvenile delinquency handling policies.]]>
Copyrights © 2026
