ABSTRACT
This chapter focuses on two problems that seek to quantify violent group behavior for the purpose of characterizing the behavioral tendencies of the group and then use these tendencies to infer the identity of the actors involved in a given violent event. It explores the group classification problem, where people seek to distil the behavior of a group into a set of features that are used to identify different groups with similar behavioral tendencies. It also talks about methodology that automates the construction of a classification system over a large number of violent groups using only open-source intelligence limited to public news reports. The open-source intelligence functions as a rich source of information about violent group behavior, and provides novel insights into the activities, organization, and behavior of violent groups. The datasets enable large-scale, quantitative analysis that can test many hypotheses about the nature and behavior of violent groups.
