Tracing Potential School Shooters in the Digital Sphere

Abstract

There are over 300 known school shooting cases in the world and over ten known cases where the perpetrator(s) have been prohibited to perform the attack at the last moment or earlier. Interesting from our point of view is that in many cases the perpetrators have expressed their views in social media or on their web page well in advance, and often also left suicide messages in blogs and other forums before their attack, along the planned date and place. This has become more common towards the end of this decennium. In some cases this has made it possible to prevent the attack. In this paper we will look at the possibilities to find commonalities of the perpetrators, beyond the fact that they are all males from eleven to roughly 25 years old, and possibilities to follow their traces in the digital sphere in order to cut the dangerous development towards an attack. Should this not be possible, then an attack should be averted before it happens. We are especially interested in the multimedia data mining methods and social network mining and analysis that can be used to detect the possible perpetrators in time. We also present in this paper a probabilistic model that can be used to evaluate the success/failure rate of the detection of the possible perpetrators.