Cyberdect. A novel approach for cyberbullying detection on twitter

Abstract

Bullying is the deliberate physical and psychological abuse that a child receives from other children. The term cyberbullying has recently emerged to denote a new type of bullying that takes place over digital platforms, where the stalkers can perform their crimes on the vulnerable victims. In severe cases, the harassment has lead the victims to the extreme causing irreparable damage or leading them to suicide. In order to stop cyberbullying, the scientific community is developing effective tools capable of detecting the harassment as soon as possible; however, these detection systems are still in an early stage and must be improved. Our contribution is CyberDect, an online-tool that seeks on Social Networks indications of harassment. In a nutshell, our proposal combines Open Source Intelligence tools with Natural Language Processing techniques to analyse posts seeking for abusive language towards the victim. The evaluation of our proposal has been performed with a case-study that consisted in monitor two real high school accounts from Spain.