MCDefender: Toward Effective Cyberbullying Defense in Mobile Online Social Networks

Abstract

Cyberbullying in Online Social Networks (OSNs) has emerged as one of the most severe social concerns. Cyberbullying can be described as a form of bullying where a perpetrator uses electronic means to cause harm to a victim. With the proliferation of smartphone technology in present times, there has been a steady shift in the usage of OSNs from traditional computers to mobile devices. However, existing systems that defend against cyberbullying are largely applicable only to traditional computing platforms and cannot be directly applied to detect cyberbullying in mobile platforms. To address such a critical issue, we investigate an innovative mobile cyberbullying defense system called MCDefender that can effectively detect and prevent cyberbullying in mobile OSNs. We first analyze the key challenges that differentiate cyberbullying conditions in traditional and mobile platforms. We then investigate a two-level detection mechanism for comprehensive cyberbullying detection in mobile OSNs where cyberbullying can be quickly detected before a cyberbullying message is sent through a mobile device and hidden cyberbullying attacks can be also detected through a more fine-grained and context-aware analysis. To demonstrate the feasibility of our approach, we implement and evaluate an Android application based on MCDefender. Our evaluation results show that our mobile application can detect cyberbullying with a high accuracy of 98.9% for OSNs.