Machine Learning Approaches To Detect And Mitigate Cyberbullying In Online Communities

Cyberbullying is one of the new problems arising from the spread of social media throughout the past several years. The best way to ensure users' safety is to aggressively search for instances of cyberbullying and stop them right away. The basis of the present research is built from sophisticated language patterns used in cyberbullying, which beyond the reach of conventional machine learning methods. This paper proposes a fresh approach based on the properties of two recurrent neural networks to detect cyberbullying incidents. One of these networks feature extracting; the other has classification ability. This method is said to enable one to solve problems. This new approach efficiently extracts residual machine learning (RML) aspects using their features. Over numerous social media sites, the findings of an experimental investigation on the classification of cyberbullying writings reveal that the RML framework much surpassed its predecessors. The model performs better in memory, accuracy, and precision than more conventional methods about the challenging issue of cyberbullying.