An Effective Cyberbullying Detection Model for the Malay Language Using Transformer Model in Social Media Platform X

In the digital era, social media has transformed communication but has also facilitated cyberbullying, significantly affecting individuals' mental health, particularly within the Malay-speaking community. Despite the growing concerns, developing effective cyberbullying detection systems for low-resource languages like Malay has been limited. This research addresses this gap by introducing transformer learning model specifically designed for detecting cyberbullying in Malay language tweets. The work begins with an extensive literature review to consolidate the current understanding of cyberbullying detection techniques. A substantial dataset will be curated from X and manually annotated, forming the basis for model training and evaluation. The research employs machine learning model and transformer models to improve overall detection accuracy and robustness. Advanced NLP techniques, including transformer models and transfer learning, will be utilized to navigate the complexities of the Malay language and ensure accurate detection. The proposed model’s performance will be rigorously evaluated using metrics such as accuracy, precision, recall, and F1-score with tests to ensure its robustness against different cyberbullying tactics. This research has produced a high-performing detection model that enhances the safety of Malay-speaking internet users and provides insights into cyberbullying dynamics within the community with advanced NLP for low-resource languages. The results demonstrate that deep learning Transformer models, particularly DeBERTa, which has outperformed traditional machine learning models in accuracy for Malay language cyberbullying detection.