Innovative Artificial Intelligence Solution as Game Changer in Cyberbullying Detection and Prevention

Abstract

The proliferation of online social networks has brought forth unprecedented connectivity and communication but has also facilitated the emergence of cyberbullying, a pervasive and harmful phenomenon. Traditional methods for identifying cyberbullying often fall short due to the dynamic nature of online interactions and the sheer volume of data. In response, this study explores the application of deep learning techniques for cyberbullying detection, focusing on the integration of LSTM networks with an attention mechanism. The research leverages a diverse dataset encompassing various forms of cyberbullying across age, ethnicity, gender, religion, and non-bullying content. Key findings reveal that the proposed models achieve high accuracy, precision, recall, and F1 scores, effectively classifying instances of cyberbullying with a comprehensive understanding of contextual nuances. Moreover, the study contributes insights into feature extraction methodologies and model optimization techniques, demonstrating the efficacy of deep learning in addressing the complexities of multi-modal social media data.