Cyberbullying Detection

Cyberbullying detection research aims to automatically identify instances of online harassment using natural language processing techniques. Current efforts focus on improving model accuracy and fairness by addressing biases in training data, exploring various deep learning architectures like transformers (e.g., BERT, RoBERTa) and recurrent neural networks (LSTMs), and developing methods for cross-platform generalization and explainability. This research is crucial for mitigating the harmful effects of cyberbullying, informing the development of effective online safety strategies, and advancing the understanding of online aggression.