Automatic Detection of Cyberbullying on Social Media

Bullying on social media is a dire problem for many youths, leading to severe health problems. In this thesis we describe the construction of a software prototype capable of automatically identifying bullying comments on the social media platform ASKfm using Natural Language Processing (NLP) and Machine Learning (ML) techniques. State of the art NLP and ML algorithms from previous research are studied and evaluated for the task of identifying bullying comments in a data set from ASKfm. The best performing classier acts as the core component in the detection software prototype. The resulting prototype can monitor selected proles on ASKfm in real time and display identied bullying comments connected to these proles on a web page