Predicting Abnormal User Behaviour Patterns in Social Media Platforms based on Process Mining

Cyberbullying has been one of the adverse repercussions of social media these days. The intensity of cyberbullying is risen considerably as a due to the increased use of image sharing and textual comments. Cyberbullying can be defined as transmitting, publishing, or circulating unbearable, harmful, false, or cruel content about some other individual. To keep the site safe and secure, automated processes for detecting certain instances are now vital. Process mining seems to be a combination of methodologies that integrate data science with management system to aid in the evaluation of organizational functions using log information. Whenever images and language that appear to be benign are combined, they might send bullying texts. As a result, different methods for analysing text and photos may fail to detect all instances of cyberbullying. In this study, we attempted to detect various examples of cyberbullying by combining textual data. The proposed system discovers hidden connections between individuals and members of a group who have similar behaviours. We achieve this through a novel strategy that incorporates techniques including data mining, analyzation of business procedures, and much more. Our study summarizes complete methodological process involved in the proposed system from the computer-generated data source in which the data pertaining user behavior, actions, and processes in an OS, software, website, or other sources to provide the visual representation of the abnormalities. Moreover, phase of identifying user behavioral patterns is the one which deserves attention.