Enhancing collaborative detection of cyberbullying behavior in Twitter data

Cyberbullying is a menace in today's socially networked world. It can have damaging physical and mental effects on the victims and hence, it needs to be tackled efficiently-several detection approaches are proposed in literature but those are mostly standalone. In this paper, we revisit the distributed and collaborative approach for detecting cyberbullying behavior using machine learning algorithms-a comprehensive enhancement of our past work-that uses many local and cloud-based collaborative configurations and different datasets. It contains a set of nodes, called detection nodes, which can identify cyberbullying employing Machine Learning classification algorithms and collaborate with each other as needed. Several experiments, consisting of various collaborative patterns, different scales, and failure scenarios, have been carried out using different Twitter(C) datasets in this study. The empirical results obtained from the experimentation show that the proposed approach is generic (i.e., allows the incorporation of different learning and collaborative techniques), and achieves better recall and precision values when compared with the stand-alone paradigm.