Abstract

Cyberbullying (CB) is one of the major cyber-issues among adolescents. In several cases, the effect of CB on victims was so severe that victims ultimately committed suicide. Despite the prevalence of automated CB detection studies using computational approaches, CB detection research still lacks empirical studies that act upon the detected CB instances in a cross-platform environment. In this paper, we propose a multi-platform, incremental self-training system that uses a decentralized learning approach to automatically detect cyberbullying instances on a minor's extended online social network. To improve the self-training model, the crowdsourced feedback of human moderators (guardians) is used. We first point out the major challenges in CB detection research and then explain our proposed design to address the discussed challenges. We conclude with the contribution to practice, and our plans for implementing the solution.

Download

Share

COinS
 

ChatterShield – A Multi-Platform Cyberbullying Detection System for Parents

Cyberbullying (CB) is one of the major cyber-issues among adolescents. In several cases, the effect of CB on victims was so severe that victims ultimately committed suicide. Despite the prevalence of automated CB detection studies using computational approaches, CB detection research still lacks empirical studies that act upon the detected CB instances in a cross-platform environment. In this paper, we propose a multi-platform, incremental self-training system that uses a decentralized learning approach to automatically detect cyberbullying instances on a minor's extended online social network. To improve the self-training model, the crowdsourced feedback of human moderators (guardians) is used. We first point out the major challenges in CB detection research and then explain our proposed design to address the discussed challenges. We conclude with the contribution to practice, and our plans for implementing the solution.