Deteksi Komentar Cyberbullying pada Media Sosial Instagram Menggunakan Algoritma Random Forest

Abstract

Cyberbullying is a phenomenon on social media where technological devices are used to insult, demean, and disrespect others. This can cause mental disorders such as loss of self-confidence, stress, depression, and even suicidal tendencies. The Ditch The Label survey, conducted by a British research institute, identified Instagram as the social media platform with the highest incidence of cyberbullying. The aim of this research is to determine the best accuracy based on the classification results of the cyberbullying comment dataset and to detect new comments as either bullying or non-bullying. One method that can be used is the random forest algorithm, which combines several similar or different methods, such as decision trees, in the classification process. The results of the classification of the testing data using the random forest algorithm show the highest accuracy of 84% in the last hyperparameter tuning combination. The built model can also detect new comments with fairly good predictive results. Suggestions for further research include classifying cyberbullying comments into more specific categories, such as racist or sexist comments.