Deep Learning for Digital Safety: Cyberbullying Detection in Arabic YouTube Content

In today's digital age YouTube is a primary platform for content sharing, communication, and entertainment. Yet, its growth has also seen the emergence of challenges like cyberbullying, which diminishes the platform's positive contributions. The challenge of addressing cyberbullying intensifies when dealing with the Arabic language, given its diverse dialects and distinct linguistic characteristics. In this study, we utilize a combination of deep learning and machine learning techniques, with a special focus on the Convolutional Long Short-Term Memory (CLSTM) model, to detect cyberbullying in Arabic YouTube comments. The primary challenge lies not only in the volume of data but also in its linguistic diversity. We curated a dataset of 365 comments from 16 YouTube videos using Python and applied the CLSTM model for analysis. Our model achieved an accuracy rate of 90%. These findings highlight the efficacy of deep learning in protecting Arabic-speaking YouTube users and present a framework for addressing similar linguistic challenges in other languages.