Risk Matrix for Violent Radicalization: A Machine Learning Approach

Abstract

Hypothesis-driven approaches identified important characteristics that differentiateviolent from non-violent radicals. However, they produced a mosaic of explanationsas they investigated a restricted number of preselected variables. Here we analyzedwithout a priory assumption all the variables of the “Profiles of Individual Radicalizationin the United States” database by a machine learning approach. Out of the 79 variablesconsidered, 19 proved critical, and predicted the emergence of violence with anaccuracy of 86.3%. Typically, violent extremists came from criminal but not radicalbackgrounds and were radicalized in late stages of their life. They were followers interrorist groups, sought training, and were radicalized by social media. They belongedto low social strata and had problematic social relations. By contrast, non-violentbut still criminal extremists were characterized by a family tradition of radicalismwithout having criminal backgrounds, belonged to higher social strata, were leadersin terrorist organizations, and backed terrorism by supporting activities. Violence wasalso promoted by anti-gay, Sunni Islam and Far Right, and hindered by Far Left, Anti-abortion, Animal Rights and Environment ideologies. Critical characteristics were usedto elaborate a risk-matrix, which may be used to predict violence risk at individual level.

Hypothesis-driven approaches identified important characteristics that differentiateviolent from non-violent radicals. However, they produced a mosaic of explanationsas they investigated a restricted number of preselected variables. Here we analyzedwithout a priory assumption all the variables of the “Profiles of Individual Radicalizationin the United States database by a machine learning approach. Out of the 79 variablesconsidered, 19 proved critical, and predicted the emergence of violence with anaccuracy of 86.3%. Typically, violent extremists came from criminal but not radicalbackgrounds and were radicalized in late stages of their life. They were followers interrorist groups, sought training, and were radicalized by social media. They belongedto low social strata and had problematic social relations. By contrast, non-violentbut still criminal extremists were characterized by a family tradition of radicalismwithout having criminal backgrounds, belonged to higher social strata, were leadersin terrorist organizations, and backed terrorism by supporting activities. Violence wasalso promoted by anti-gay, Sunni Islam and Far Right, and hindered by Far Left, Anti-abortion, Animal Rights and Environment ideologies. Critical characteristics were usedto elaborate a risk-matrix, which may be used to predict violence risk at individual level.