Building Prediction Model for Detecting Cyberbullying using TikTok Comments

Easily accessible Internet and the need for communication led to the widespread use of social media platforms, e.g., TikTok. Social media platforms offer opportunities for social interaction and enter- tainment but also introduce risks such as fake news, fraud, and cyberbullying. The latter severely threatens the victim’s mental health and overall well-being. This research collects, annotates, and analyzes 1,508 TikTok comments concerning cyberbullying behavior. Then, the comments were used to build a Deep Learning BERT (Bidirectional Encoder Representations from Transformers) architecture prediction model. This experiment will use pretrained model BERT and fine-tuning BERT. After fine-tuning, the prediction model achieves 0.63 validation accuracy. This research high- lights the importance of building a prediction model to detect cyberbullying and contribute to a better understanding and prevention of this behavior on social media platforms.