基于社会网络与事件关联的恐怖事件监测与识别  

Terrorist events monitoring and identifying based on correlation between social networks and events

作　　者：李泽[1] 孙多勇[1] 李博[1] 

机构地区：[1]国防科技大学信息系统与管理学院,长沙410073

出　　处：《科技导报》2017年第9期87-94,共8页Science & Technology Review

基　　金：国家自然科学基金项目(71473263);高等学校博士学科专项科研基金项目(20134307110020)

摘　　要：恐怖组织的社会网络结构变化与恐怖事件的发生具有一定的关联性。基于此关联,通过监测恐怖组织社会网络的变化,可以实时、有效地识别恐怖事件。将基于社会网络变化检测的恐怖事件监测与识别问题视为分类问题,并通过神经网络模型进行分类研究。以某一时刻是否发生恐怖事件为标准,对恐怖组织社会网络进行分类;通过网络分析技术,得出网络的参数指标,建立混合算法改进的神经网络模型;将网络的参数指标与恐怖事件发生情况分别作为输入和输出,对神经网络进行训练与测试。案例分析和对比结果表明,基于神经网络模型的社会网络变化检测方法具备较好的恐怖事件监测与识别能力;该方法可在一定程度上弥补现有方法正确率不高、通用性不强、检测结果与恐怖事件实际发生的相关性不高等不足。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.

关 键 词：恐怖组织网络 变化检测 恐怖事件 监测与识别 神经网络 

分 类 号：C912.3[经济管理]