The social networks of terrorist organization and terrorist events are changing correlatively. Based on the correlation between social networks and events, detecting changes in the networks may effectively help monitor and identify terrorist events. Terrorist attack early-warning is regarded as a classification problem, and neural network is used to solve this problem. The time when any terrorist attack happens is identified as a "change" point. Then the corresponding network is labeled as a changed one. Accordingly, the time sequence networks are classified into two sets:"changed" and "unchanged". Measures of networks are obtained by social network analysis to represent networks. Hybrid heuristic algorithms are applied to optimizing the neural network. The classified network measures and the Boolean data of whether the terrorist events have happened are taken as the input and output, respectively. A real-world case study is given to show that detecting changes in terrorist networks based on neural network has the ability to monitor and identify terrorist events. Comparison results also show that the proposed approach can solve the problems such as versatility, accuracy and correlation encountered by the existing methods to some extent.

terrorist network  /  change detection  /  terrorist events  /  monitoring and identifying  /  neural network