Cyberbullying System Detection and Analysis

Abstract

Abstract-Cyber-bullying has recently been reported as one that causes tremendous damage to society and economy. Advances in technology related to web-document annotation and the multiplicity of the online communities renders the detection and monitoring of such cases rather difficult and very challenging. This paper describes an online system for automatic detection and monitoring of Cyber-bullying cases from online forums and online communities. The system relies on the detection of three basic natural language components corresponding to Insults, Swears and Second Person. A classification system and ontology like reasoning have been employed to detect the occurrence of such entities in the forum / web documents, which would trigger a message to security in order to take appropriate action. The system has been tested on two distinct forums and achieves reasonable detection performances