An Automatic Method to Prevent and Classify Cyber Bullying Incidents Using Machine Learning Approach

Deep learning models for CyberBullying detection on social media, as a relatively new field of research and application, need the identification, investigation, and analysis of a broad range of human-based expressions. Natural language processing encounters a categorization challenge when trying to identify whether a remark, post, message, or picture represents bullying. A thorough investigation of the meaning of words is also required. Previous attempts to identify CyberBullying focused mostly on manual feature extraction approaches. These tactics not only take a lot of time and effort, but also often misread the intended meaning of a communication. This removes the need for any further feature extraction approaches. Deep learning was used to identify CyberBullying in social media data that included both textual and visual aspects. Because of the large volume and diversity of user-generated information on current social media platforms, detecting CyberBullying in real time has become more challenging. The rapid transmission of information makes real-time regulation of online speech problematic. CyberBullying is a widespread issue that may take many forms. To stop CyberBullying, researchers used deep learning models to analyse social media content, modality, and language. This research found that embeddings with deep learning architectures accelerate representation learning and feature selection compared to typical machine learning methods.