Abstract
This study presents a deep learning-based approach to detecting cyberbullying in Kazakh, addressing key challenges in low-resource languages. The research highlights the increasing prevalence of cyberbullying in Kazakhstan, the limitations of traditional machine learning models, and the need for advanced text classification techniques. A novel hybrid deep learning model integrating Convolutional Neural Networks (CNNs), Bidirectional Long Short-Term Memory (BiLSTM), and Transformer-based architectures is proposed to enhance detection accuracy. The study outlines the dataset collection process, data augmentation techniques, and model evaluation using key performance metrics, including accuracy, precision, recall, and F1-score. Experimental results demonstrate that the proposed model significantly outperforms conventional machine learning algorithms and previously published methods. The findings offer practical implications for automated content moderation on social media platforms and contribute to advancing natural language processing (NLP) tools for the Kazakh language. This study proposes a novel hybrid deep learning model for Kazakh-language cyberbullying detection. The model integrates CNN, BiLSTM, and Transformer-based architectures to address the challenges of low-resource languages, complex morphology, and evolving online slang. Experimental results demonstrate that the proposed approach significantly outperforms traditional machine learning algorithms and previously published methods. The research contributes to the development of NLP tools for Kazakh and provides practical solutions for automated content moderation on social media platforms.