Complex networks for terrorist target prediction

Developments in statistics and computer science have influenced research on many social problems. This process also applies to the study of terrorism. In this context, network analysis is one of the most popular mathematical methods for analyzing terrorist organizations and dynamics. Nonetheless, few studies have applied network science to the analysis of terrorist events. Therefore, in this work we first introduce a novel method to analyze the heterogeneous dynamics of terrorist attacks through the creation of a dynamic meta-network of terror for the period 1997–2016. Second, we use our terrorist meta-network to test the power of Network-based Inference algorithm in predicting terrorist targets. Results are promising and show how this algorithm reaches high levels of precision, accuracy, and recall and indicate that network outcomes can be used in broader machine learning models.