Detection of Cyberbullying Using Machine Learning-Based Model
Abstract
Cyberbullying is a serious and widespread issue that affects lots of people around
the world. Cyberbullying can have negative consequences for the online platforms,
such as reduced user engagement, trust, and satisfaction, and increased legal and
reputational risks. This study leverages advanced machine learning to detect
cyberbullying with high accuracy and scalability. The study developed an API for
real-time monitoring and intervention capabilities that allow immediate action
against cyberbullying incidents, preventing escalation and mitigating harm. A
variety of machine learning algorithms and a type of deep learning algorithm were
investigated for their efficacy in detecting cyberbullying. These included Support
Vector Machines (SVMs), Random Forests, Logistic Regression, and
Convolutional Neural Networks (CNNs). The result showed high accuracy with
balanced precision and recall using machine learning.