The purpose of this study is to compare the effectiveness of deep learning methods designed to automatically detect signs of cyberbullying in textual data from digital communications and to identify the most effective neural network architecture. Cyberbullying is one of the most pressing socio-psychological problems of the information society and negatively affects users’ psychological and emotional well-being, particularly that of adolescents and young people. The increasing volume of textual content on social networks and online platforms requires the development of automated systems capable of promptly detecting messages containing insults, aggression, and threats.

The study examined a multilayer perceptron (MLP), a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory network (LSTM), and a deep neural model with an attention mechanism using a corpus of labeled texts. Model performance was evaluated using accuracy, recall, precision, and the F1-score. The results showed that the MLP model has limited ability to capture complex contextual relationships, the CNN architecture effectively identifies local linguistic patterns, and the RNN and LSTM models are better at processing word sequences and long-term dependencies. The attention-based model demonstrated the highest classification performance because it assigns greater importance to significant text fragments containing signs of aggression, intimidation, and abuse.

The findings showed that model effectiveness depends on the quality and balance of labeled data. Text preprocessing, word normalization, and tokenization contributed to improved training performance. To apply the proposed models in real-world online environments, they must be regularly retrained using new data. Adapting the algorithms to the morphological and semantic features of the Kazakh language will improve the monitoring of Kazakhstan’s digital environment. Future research will employ transformer-based architectures, multilingual datasets, and explainable artificial intelligence methods. The findings can be used to improve systems for the early detection and prevention of cyberbullying.

ABDRAKHMANOV R.B.

Candidate of technical sciences, associate professor, International university of tourism and hospitality, Turkistan, Kazakhstan

E-mail: abdrakhmanov.rustam@iuth.edu.kz, https://orcid.org/0000-0002-5508-389X

KERIMKHAN B.T.

PhD, senior lecturer at the department of computer and software engineering, L.N. Gumilyov Eurasian national university, Astana, Kazakhstan

E-mail: bek_zhan_16@mail.ru, https://orcid.org/0000-0002-8480-7316

KASYMBEKOV A.S.

Candidate of technical sciences, associate professor, International university of tourism and hospitality, Turkistan, Kazakhstan

E-mail: abbazbek@mail.ru, https://orcid.org/0000-0001-8022-5637

ISKAKOV T.B.

Candidate of technical sciences, associate professor, International university of tourism and hospitality, Turkistan, Kazakhstan

E-mail: abbazbek@mail.ru, https://orcid.org/0000-0001-8022-5637

UKSIKBAYEV E.Z.

Master’s degree holder, Khoja Akhmet Yassawi international kazakh-turkish university, Turkistan, Kazakhstan

E-mail: erkebulan.uxikbayev@ayu.edu.kz, https://orcid.org/0009-0003-3868-6773

1. Alsubait T., Alfageh D. Comparison of machine learning techniques for cyberbullying detection on YouTube Arabic comments // International Journal of Computer Science and Network Security, — 2021. — Vol. 21, no. 1. — P. 1–5. — URL: https://koreascience.kr/article/JAKO202121055656002.page
2. Dewani M., Memon M., Bhatti, S. Cyberbullying detection: Advanced preprocessing techniques & deep learning architecture for Roman Urdu data. Journal of Big Data, — 2021. — Vol. 8. — Art. 160. — DOI: 10.1186/s40537-021-00550-7.
3. Hall D., Silva Y., Wheeler Y., Cheng L., Baumel, K. Harnessing the power of interdisciplinary research with psychology-informed cyberbullying detection models. // International Journal of Bullying Prevention. — 2022. — Vol. 4, no. 1. — P. 47–54. — DOI: 10.1007/s42380-021-00107-5.
4. Arce-Ruelas K. I., Alvarez-Xochihua O., Pellegrin L., Cardoza-Avendaño L., González-Fraga J. Á. // Automatic cyberbullying detection: A Mexican case in high school and higher education students. IEEE Latin America Transactions. — 2022. — Vol. 20, no. 5. — P. 770–779. — URL: https://latamt.ieeer9.org/index.php/transactions/article/view/5934
5. Ahmed T., Rahman M., Nur S., Islam, A., Das, D. Natural language processing and machine learning based cyberbullying detection for Bangla and romanized Bangla texts. // TELKOMNIKA (Telecommunication Computing Electronics and Control). — 2022. — Vol. 20, no. 1. — P. 89–97. — DOI: 10.12928/TELKOMNIKA.v20i1.18630.
6. Omarov A., Tursynova A., Postolache O., Gamry K., Batyrbekov, A., et al. Modified UNet model for brain stroke lesion segmentation on computed tomography images. // Computers, Materials & Continua. — 2022. — Vol. 71, no. 3. — P. 4701–4717. — DOI: 10.32604/cmc.2022.020998.
7. Al-Marghilani. Artificial intelligence-enabled cyberbullying-free online social networks in smart cities. // International Journal of Computational Intelligence Systems. — 2022. — Vol. 15. — Art. 9. — DOI: 10.1007/s44196-022-00063-y.
8. Theng C., Othman N., Abdullah R., Anawar S., Ayop Z., et al. Cyberbullying detection in Twitter using sentiment analysis. // International Journal of Computer Science and Network Security. — 2021. — Vol. 21, no. 11. — P. 1–10. — URL: https://koreascience.kr/article/JAKO202105254945209.do
9. Sadiq S., Mehmood A., Ullah S., Ahmad M., Choi G., et al. Aggression detection through deep neural model on Twitter. // Future Generation Computer Systems. — 2021. — Vol. 114. — P. 120–129.
10. Sarac Essiz E., Oturakci M. Artificial bee colony-based feature selection algorithm for cyberbullying. // The Computer Journal. — 2021. — Vol. 64, no. 3. — P. 305–313. — DOI: 10.1093/comjnl/bxaa066.
11. Gomez C. E., Sztainberg M. O., Trana R. E. Curating cyberbullying datasets: A human-AI collaborative approach. // International Journal of Bullying Prevention. — 2022. — Vol. 4, no. 1. — P. 35–46. — DOI: 10.1007/s42380-021-00114-6.
12. Salawu S., Lumsden J., He Y. A mobile-based system for preventing online abuse and cyberbullying. // International Journal of Bullying Prevention. — 2022. — Vol. 4, no. 1. — P. 66–88. — DOI: 10.1007/s42380-021-00115-5.
13. Mahajan E., Mahajan H., Kumar S. EnsMulHateCyb: Multilingual hate speech and cyberbully detection in online social media. // Expert Systems with Applications. — 2024. — Vol. 236. — Art. 121228. — DOI: 10.1016/j.eswa.2023.121228.
14. Jamjoom A. A., Karamti H., Umer M., Alsubai S., Kim T. H., Ashraf, I. Robertanet: Enhanced roberta transformer based model for cyberbullying detection with glove features. // IEEE Access. — 2024. — Vol. 12. — P. 58950–58959. — DOI: 10.1109/ACCESS.2024.3386637.
15. Zhang J., Wu Q., Xu Y., Cao C., Du Z., Psounis, K. Efficient toxic content detection by bootstrapping and distilling large language models. // Proceedings of the AAAI Conference on Artificial Intelligence. — 2024. — Vol. 38, no. 19. — P. 21779–21787. — DOI: 10.1609/aaai.v38i19.30178.
16. Xu M., Liu S. Rb_bg_mha: A roberta-based model with bi-gru and multi-head attention for chinese offensive language detection in social media. // Applied Sciences. — 2023. — Vol. 13, no. 19. — Art. 11000. — DOI: 10.3390/app131911000.

cyberbullying, deep learning, neural networks, natural language processing, text classification, attention mechanism