Nature-Inspired-Based Approach for Automated Cyberbullying Classification on Multimedia Social Networking

In the modern era, the cyberbullying (CB) is an intentional and aggressive action of an individual or a group against a victim via electronic media. The consequence of CB is increasing alarmingly, affecting the victim either physically or psychologically. This allows the use of automated detection tools, but research on such automated tools is limited due to poor datasets or elimination of wide features during the CB detection. In this paper, an integrated model is proposed that combines both the feature extraction engine and classification engine from the input raw text datasets from a social media engine. The feature extraction engine extracts the psychological features, user comments, and the context into consideration for CB detection. The classification engine using artificial neural network (ANN) classifies the results, and it is provided with an evaluation system that either rewards or penalizes the classified output. The evaluation is carried out using Deep Reinforcement Learning (DRL) that improves the performance of classification. The simulation is carried out to validate the efficacy of the ANN-DRL model against various metrics that include accuracy, precision, recall, and f-measure. The results of the simulation show that the ANN-DRL has higher classification results than conventional machine learning classifiers.