Media of University of Babylon | Master’s Thesis Defense at the College of Information Technology on a Hybrid Model for Cyberbullying Detection Using Deep Learning Techniques

Media of University of Babylon - كلية الطب

Master’s Thesis Defense at the College of Information Technology on a Hybrid Model for Cyberbullying Detection Using Deep Learning Techniques

by : Duhaa Fadill Abbas

Date : 20/2/2026 27:27:10 PM

Veiw : 169

Master’s Thesis Defense at the College of Information Technology on a Hybrid Model for Cyberbullying Detection Using Deep Learning Techniques

Duha Fadill Abbas
The Department of Software at the College of Information Technology, University of Babylon, witnessed the defense of the Master’s thesis submitted by Ali Jasim Mohammed, entitled “A Hybrid Model for Cyberbullying Detection Using BERT CNN and Multi Head Attention,” under the supervision of Prof. Dr. Wafaa Mohammed Saeed. The defense was conducted in a rigorous academic atmosphere that reflected the significance of the research at the intersection of artificial intelligence, natural language processing, and digital content analysis, and its role in addressing the growing challenges within modern online communication environments.

The thesis highlighted cyberbullying as one of the most prominent forms of digital abuse in the era of social media, emphasizing its deep psychological, health, and social impacts on individuals, particularly with the increasing reliance on digital platforms. The study pointed out that although eliminating this phenomenon entirely remains difficult, it is still possible to mitigate its effects through the development of intelligent systems capable of early detection and accurate analysis of harmful content.

The researcher proposed a hybrid model named BERT CNN Multi head Attention, which combines Bidirectional Encoder Representations from Transformers BERT for extracting contextual meaning from text, Convolutional Neural Networks CNN for local feature extraction, and Multi head Attention mechanisms for capturing deep semantic patterns. The design aims to leverage the strengths of each component in order to enhance detection performance and improve the classification accuracy of multi class cyberbullying cases.

The study concluded that the proposed hybrid model represents an important step toward developing intelligent deep learning based solutions for addressing digital abuse issues, contributing to safer online communication environments by supporting early detection mechanisms and reducing the spread of harmful content.

ت	اسماء اعضاء لجنة المناقشة	اللقب العلمي	الاختصاص الدقيق	مكان العمل	المنصب
1	د. ايمان صالح صكبان	استاذ	ذكاء اصطناعي و معلوماتية حيوية	جامعة بابل / كلية تكنولوجيا المعلومات	رئيساً
2	د. ايناس فاضل عبدالله	استاذ مساعد	ذكاء اصطناعي و معلوماتية حيوية	جامعة الكوفة / كلية التربية للبنات	عضوا
3	د.حسين علي اسماعيل	مدرس	ذكاء اصطناعي	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا
4	د. وفاء محمد سعيد	استاذ	ذكاء اصطناعي و معالجة صورية	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا و مشرفا