Identifying The Responsible Group for Extreme Acts of Violence Through Pattern Recognition

The expansion of Internet has eased the broadcasting of data, information, and propaganda. The availability of myriads of social and televised media have turned the spotlight on violent extremism, widened the rift between different sides of the spectrum, and expanded the scope and impact of ideology-oriented acts of violence on citizens and nations. The human casualties and psychological impacts on societies make any study on such acts worthwhile, let alone attempting to detect patterns among them. This study focuses on mining the information about each violent act, including human casualties and fatalities, level of coordination and expertise, importance of the targeted process, and the extent of its impact on the process, to identify the responsible group. Decision tree, a non-linear classifier, reached 20% cross-validation accuracy in identifying the correct group among 38 groups. This is the highest accuracy achieved in comparison with other linear classifiers, including Perceptron, SVM, and least squares. Our results also underscored the human casualties and fatalities as the most important predictors. The other four variables, including level of coordination, level of expertise, importance of the targeted process, and the extent of the impact on the process were all partly correlated and less helpful. However, the single feature, generated by linear combination of these four features using PCA, was as good of a predictor as the human casualties and fatalities.