Graph-Based Representation and Detection of Cyberbullying on Social Media Using GCNs

The widespread use of social media has contributed to an increase in instances of cyberbullying, which constitutes hateful and offensive interactions between users of digital platforms, necessitating effective detection methods to mitigate its impact. This study presents a novel approach to cyberbullying detection by employing Graph Convolutional Networks (GCNs), which are adept at processing graph-structured data, used to represent various complex relationships between words in text data. Trained on a substantial dataset of cyberbullying tweets, the architecture of the proposed GCN allows for the aggregation of features based on the relationships between tweets and words, enabling the model to capture complex dependencies inherent in the data. The model achieved a significant accuracy of $93.1 \%$, demonstrating high precision and recall metrics. This work not only advances the state of the art in automated cyberbullying detection but also lays the groundwork for future research leveraging graph-based methodologies in natural language processing tasks.