Terrorist Group Behavior Prediction by Wavelet Transform-Based Pattern Recognition

Predicting terrorist attacks by group networks is an important but difficult issue in intelligence and security informatics. Effective prediction of the behavior not only facilitates the understanding of the dynamics of organizational behaviors but also supports homeland security’s missions in prevention, preparedness, and response to terrorist acts. There are certain dynamic characteristics of terrorist groups, such as periodic features and correlations between the behavior and the network. In this paper, we propose a comprehensive framework that combines social network analysis, wavelet transform, and the pattern recognition approach to investigate the dynamics and eventually predict the attack behavior of terrorist group. Our ideas rely on social network analysis to model the terrorist group and extract relevant features for group behaviors. Next, based on wavelet transform, the group networks (features) are predicted and mutually checked from two aspects. Finally, based on the predicted network, the behavior of the group is recognized based on the correlation between the network and behavior. The Al-Qaeda data are investigated with the proposed framework to show the strength of our approaches. The results show that the proposed framework is highly accurate and is of practical value in predicting the behavior of terrorist groups.