A Multichannel Deep Learning Framework for Cyberbullying Detection on Social Media

Online social networks (OSNs) play an integral role in facilitating interaction; however, these increase antisocial behavior, such as cyberbullying, hate speech, and trolling. Aggression or speech that takes place through short message service (SMS) the Internet (e.g., media platforms) is known cyberbullying. Therefore, automatic detection utilizing natural language processing (NLP) a necessary first step helps prevent This research proposes cyberbullying method to detect aggressive behavior using consolidated deep learning model. technique utilizes multichannel based on three models, namely, bidirectional gated recurrent unit (BiGRU), transformer block, convolutional neural network (CNN), classify Twitter comments into two categories: not aggressive. Three well-known datasets were combined evaluate performance of proposed method. The achieved promising results. accuracy was approximately 88%.