Comparative Analysis of Machine Learning and Deep Learning Techniques for Predicting and Detecting Cyberbullying on Social Media

This study investigates the effectiveness of various machine learning and deep learning algorithms in predicting and detecting cyberbullying incidents on social media platforms. Utilizing the comprehensive CB_Label dataset, which captures key aspects of cyberbullying, we assess multiple performance metrics, including accuracy, precision, recall, F -measure, and execution time. Our analysis reveals notable variations in performance across different algorithms, including Naive Bayes, Bayes Net, Logistic Regression, and Hoeffding Tree, among others. While some algorithms demonstrate high accuracy and precision, others exhibit significant differences in their ability to classify instances correctly and minimize errors. The results indicate that machine learning and deep learning techniques do not exhibit uniform effectiveness, leading to the rejection of the hypothesis that no significant differences exist between them. This finding emphasizes the importance of selecting appropriate algorithms based on specific performance measures when addressing the complex challenge of cyberbullying detection in real -world applications. The study contributes to the growing body of research on cyberbullying detection and provides insights into optimizing algorithmic choices for effective intervention strategies.