Denigration Bullying Resolution using Wolf Search Optimized Online Reputation Rumour Detection

Abstract

Denigration is the most common bullying tactic involving public figures like celebrities and politicians where rumourous stories, pictures and videos are posted online to discredit and defame. It involves online “dissing” or “gossiping” about someone by writing and distributing vulgar, derogatory, cruel, mean, or untrue rumours. An online denigrate comment is typically posted as a malicious viral rumour to hurt the victim. A model to detect defamatory posts in the form of online reputation rumours can facilitate pinpointing cases of denigration in target profiles. The key bottlenecks in analyzing rumours in real-time are characterized by the cross-platform, cross-lingual, multimodal, often skewed, high-dimensional nature of data. Optimal feature selection can avoid the curse of dimensionality, increase model accuracy, decrease model training time and enhance the generalizability of the model by reducing overfitting. This research proffers an implicit mechanism that comprehends the truth value of the reputation rumour using meta-heuristic optimization algorithm, Wolf Search algorithm (WSA). An empirical evaluation on the rumour dataset affirms that using feature selection maximizes the relevance and minimizes the redundancy in feature set to build an efficient rumour classification model.