Benchmarking Multiple Machine Learning Algorithms for Sentiment Analysis on Sexual Violence

Sexual violence is not a new phenomenon in Indonesian society. It has occurred among women, men, and adolescents in various global conflicts. YouTube comments are a good source of data for sentiment analysis because they provide a large amount of text-based input from users. The data used in this study were collected from YouTube comments through a crawling process conducted on April 3, 2024. From crawling the comments on 6 videos, the data were combined into a single CSV file with a total of 11,628 entries, which was reduced to 11,133 entries after the cleansing stage. The results of this study show that the SVM model achieved the highest accuracy at 82.50%, followed by Logistic Regression with an accuracy of 82.18%. The Random Forest and XGBoost models had almost identical accuracies, at 81.64% and 81.60% respectively. The Naive Bayes model had an accuracy of 80.11%, while KNN showed the lowest accuracy among all models, at 78.39%.