PhD, University of Cincinnati, 2024, Education, Criminal Justice, and Human Services: Information Technology
Over the past ten years, cyberbullying has become a prevalent issue across various
levels of education and society globally. This dissertation delves into the complex
landscape of cyberbullying text detection. Through a thorough parametric analysis, it
explores the intricacies of cyberbullying text detection research, presenting insights into
potential solutions and strategies. A case study is conducted to investigate cultural
variations and perceptions of offensiveness, particularly within Ghanaian culture,
contributing to a deeper understanding of cyberbullying dynamics. The dissertation also
explores strategies for prevention and fostering a safer online environment, along with
examining cultural interpretations of technology features. Furthermore, this dissertation
focuses on detecting cyberbullying in adversarial text content within social networking
site, with a specific emphasis on identifying hate speech. Utilizing a deep learning-based
approach with a correction algorithm, this dissertation yielded significant results. An
LSTM model with a fixed epoch of 100 demonstrated remarkable performance, achieving
high accuracy, precision, recall, F1-score, and AUC-ROC scores of 87.57%, 88.73%,
87.57%, 88.17%, and 91% respectively. The LSTM model's performance surpassed that
of previous studies when compared. Additionally, the dissertation offers recommendations
for defense strategies against adversarial attacks on AI-based models, providing valuable
insights for future research endeavors.
Committee: Nelly Elsayed Ph.D. (Committee Chair); Amanda La Guardia Ph.D. (Committee Member); Zaghloul Elsayed Ph.D. (Committee Member); M. Murat Ozer Ph.D. (Committee Member)
Subjects: Information Technology