Abstract

Cyberbullying is a major problem encounteredon internet that affects teenagers and also adults.It has lead tomishappenings like suicide and depression.Regulation ofcontent on Social media platorms has become a growing need.The following study uses data from two different forms ofcyberbullying, hate speech tweets from Twittter and commentsbased on personal attacks from Wikipedia forums to build amodel based on detection of Cyberbullying in text data usingNatural Language Processing and Machine learning. Threemethods for Feature extraction and four classifiers are studiedto outline the best approach. For Tweet data the modelprovides accuracies above 90% and for Wikipedia data it givesaccuracies above 80%.