Klasifikasi Komentar Mengandung Body Shaming pada Media Sosial dengan Menggunakan Algoritma Convolutional Neural Network

Abstract

The use of social media, which can be easily accessed without limits, can contribute to online bullying, namely body shaming. In social media, this action can be found in comments, tweets, or contents. Body shaming on social media can be in the form of negative criticism of the overall body shape to others. This causes concern and annoyance to the victimized social media users. The large number of body shaming comments that can be found makes it difficult to classify body shaming comments manually by professional. Therefore, with the development of information technology today, a system can be designed that can classify comments containing body shaming elements more effectively. This research aims to classify comments containing body shaming on social media by applying the Convolutional Neural Network algorithm and IndoBERT as word embedding. The data used are 3500 Indonesian comments crawled on Instagram and TikTok social media applications which include body shaming comments and comments that do not include body shaming. The results of the model performance evaluation through confusion matrix obtained an accuracy of 89%. Based on the evaluation results, it shows that the developed system is able to classify body shaming comments with sufficient accuracy.