Abstract Anti-terrorism research is an important academic topic in current societies. The crucial features of attacked incidents can be obtained effectively by identifying phase division of terrorism history. To handle time-series issues, complex networks theories are efficient and reliable analysis solutions. Therefore, we propose an original community detection method for complex time-series networks. Especially, we consider the improved local density operator and bi-directional neighbor retrieval (ILD-BNR). First, complex networks of threatened countries are established by incidents feature and time-series principles. Then, cores of networks are selected by improved density operator. After that, attributes of unstable nodes are revised iteratively until initialization is finished. The optimal classification results are obtained by retrieval pattern of bi-directional neighbor. Finally, on the basis of clustering consequences, historical phases are divided ultimately. The mechanism of each phase is discussed simultaneously. The experiments demonstrate some important conclusions: a) The accuracy of proposed method is better than other evaluated algorithms on real time-series networks; b) The historical phase number is reduced reasonably, which is beneficial to analysis of information; and c) Classification consequences can reflect the historical tendency of terrorism.

中文翻译：

---

恐怖主义历史阶段划分研究：一种基于时间序列复杂网络的分析方法

摘要 反恐研究是当今社会的一个重要学术课题。通过识别恐怖主义历史的阶段划分，可以有效地获取袭击事件的关键特征。为了处理时间序列问题，复杂网络理论是有效且可靠的分析解决方案。因此，我们提出了一种用于复杂时间序列网络的原始社区检测方法。特别是，我们考虑了改进的局部密度算子和双向邻居检索（ILD-BNR）。首先，受威胁国家的复杂网络是通过事件特征和时间序列原则建立的。然后，通过改进的密度算子选择网络的核心。之后，迭代修改不稳定节点的属性，直到初始化完成。通过双向邻域检索模式得到最优分类结果。最后，在聚类结果的基础上，最终划分历史阶段。同时讨论每个阶段的机制。实验证明了一些重要的结论： a) 在实时序列网络上，所提出的方法的准确性优于其他评估算法；b) 合理减少历史阶段数，有利于信息分析；c) 分类后果可以反映恐怖主义的历史趋势。a) 在实时序列网络上，所提出方法的准确性优于其他评估算法；b) 合理减少历史阶段数，有利于信息分析；c) 分类后果可以反映恐怖主义的历史趋势。a) 在实时序列网络上，所提出方法的准确性优于其他评估算法；b) 合理减少历史阶段数，有利于信息分析；c) 分类后果可以反映恐怖主义的历史趋势。