Abstract
Cyber extremism has become a major predicament in recent years, increasing the amount of research being conducted on it. In this work, we propose a three-staged data and social network-oriented approach to classify videos on YouTube and identify cyber extremism hotspots and visualise their emergence over the years. The first stage consists of building up a corpus by using tweets and audio clips of extremist groups and refining it further using tf-idf. Second stage involves searching extremist videos on Y ouTube with the help of bigrams developed from the corpus made in the previous stage. At the last stage, these videos are manually tagged and later, classified and clustered using Naive Bayes classifier and hierarchical clustering. Finally, locations from the thus extremist labelled videos are identified.