A Contextual Transformer Architecture for Proactive Recognition of Cyber Aggression in Social-Media Communication

This paper presented a contextual transformer architecture for identifying cyber aggression in social-media text. The proposed formulation divides messages into non-aggressive, explicit-aggression, and implicit-aggression categories and employs a fine-tuned BERT encoder to capture contextual relationships. According to the reported experimental values, the BERT model obtains the strongest performance among the evaluated BERT, Bi-LSTM, and SVM approaches. An important aspect of the framework is the separate treatment of implicit aggression. Hostile intent is not always represented by an obvious offensive word; it can emerge from the relationship between otherwise ordinary terms, sentence structure, or contextual cues. The transformer-based represen-tation is therefore appropriate for investigating such cases. Nevertheless, practical deployment must address limitations related to language diversity, evolving slang, platform differ-ences, computation, privacy, fairness, and model interpretation. Future development can extend the framework toward mul-tilingual and code-mixed communication, conversational con-text, and multimodal content. Images, audio, and video could be incorporated when textual evidence alone is insufficient. User-level and conversation-level signals could provide ad-ditional context while requiring careful privacy controls. A future implementation could also provide confidence scores and human-readable explanations so that automated predic-tions support rather than replace moderator decisions. Finally, the trained model could be integrated into a scalable real-time service for early screening and prioritization of potentially harmful content.

[1]     C. Raj, A. Agarwal, G. Bharathy, B. Narayan, and M. Prasad, “Cy-berbullying detection: Hybrid models based on machine learning and natural language processing techniques,” Electronics, vol. 10, no. 22, p. 2810, 2021. [2]     M. H. U. Rahman, M. Divya, B. R. Reddy, K. S. Kumar, and P. R. Vani, “Cyberbullying detection using LSTM networks on social media text,” International Journal of Research in Applied Science and Engineering Technology, vol. 10, 2022. [3]     J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186. [4]     M. Zampieri, S. Malmasi, P. Nakov, S. Rosenthal, N. Farra, and R. Kumar, “Predicting the type and target of offensive posts in social media,” in Proc. NAACL-HLT, 2019, pp. 1415–1420. [5]     D. Kiela, H. Firooz, A. Mohan, V. Goswami, A. Singh, P. Ringshia, and D. Testuggine, “The hateful memes challenge: Detecting hate speech in multimodal memes,” in Advances in Neural Information Processing Systems, vol. 33, 2020. [6]     E. Wulczyn, N. Thain, and L. Dixon, “Ex Machina: Personal attacks seen at scale,” in Proc. 26th International Conference on World Wide Web, 2017, pp. 1391–1400. [7]     A. Schmidt and M. Wiegand, “A survey on hate speech detection using natural language processing,” in Proc. 5th International Workshop on Natural Language Processing for Social Media, 2017, pp. 1–10. [8]     P. Fortuna and S. Nunes, “A survey on automatic detection of hate speech in text,” ACM Computing Surveys, vol. 51, no. 4, pp. 1–30, 2018. [9]     P. Badjatiya, M. Gupta, M. Gupta, and V. Varma, “Deep learning for hate speech detection in tweets,” in Proc. 26th International Conference on World Wide Web Companion, 2017, pp. 759–760. [10]    S. Agrawal and A. Awekar, “Deep learning for detecting cyberbullying across multiple social media platforms,” in Advances in Information Retrieval, Springer, 2018, pp. 141–153.