For many decades, the contagion of terrorism has presented a huge risk to our global society. To this end, previous research has found that single terrorist attacks can elevate the risk of subsequent attacks nearby, an idea referred to as the phenomenon. A near-repeat pair consists of two attacks that occur within a specific time period and spatial distance of one another (e.g, one week and 10 miles), and the existence of near-repeats reflects calculated decisions by terrorists as they plan when and where to attack. Thus, enhancing our understanding of these attack patterns can shed light into the operational strategies of terrorist organizations. To this end, there remains key gaps in current knowledge regarding whether near-repeat attack patterns generalize across major terrorist organizations (i.e., do all major organizations exhibit near-repeat activity?), and the major risk factors associated with near-repeat terrorism (e.g., attack tactics). Utilizing data on over 50,000 terrorist attacks, this study seeks to fill these gaps in knowledge by first analyzing near-repeat attack patterns both across and within major terrorist organizations, and subsequently developing a statistical learning pipeline to identify the risk factors which are most salient in near-repeat attacks. We find that near-repeat terrorism occurs at a statistically significant level for 28 out of the 30 organizations studied. Although our findings show that near-repeat attacks are common amongst organizations, we find that the tactics (e.g., weapon choice, target type) utilized within near-repeat attacks vary vastly between organizations. The results of this work offer insights that augment our knowledge of terrorism patterns, which can in turn improve counterterrorism operations. Further, our end-to-end data-driven framework offers a strong decision support tool in which users can first detect near-repeat attacks, and subsequently identify the key operational patterns within those attacks (for any terrorist organizations of interest).

中文翻译：

---

恐怖重演：重大恐怖组织时空攻击模式识别与分析

几十年来，恐怖主义的蔓延给我们的全球社会带来了巨大的风险。为此，之前的研究发现，单次恐怖袭击会提高附近发生后续袭击的风险，这种想法被称为“现象”。接近重复对由在特定时间段和彼此空间距离（例如，一周和 10 英里）内发生的两次袭击组成，并且接近重复的存在反映了恐怖分子在计划何时何地时经过深思熟虑的决定去攻击。因此，加强我们对这些攻击模式的理解可以揭示恐怖组织的行动策略。为此，目前关于近乎重复的攻击模式是否在主要恐怖组织中普遍存在（即，是否所有主要组织都表现出近乎重复的活动？）以及与近乎重复的恐怖主义相关的主要风险因素，目前的知识仍然存在重大差距（例如，攻击策略）。本研究利用超过 50,000 起恐怖袭击的数据，首先分析主要恐怖组织之间和内部的近乎重复的袭击模式，然后开发统计学习渠道，以确定最突出的风险因素，从而填补这些知识空白。近乎重复的攻击。我们发现，在所研究的 30 个组织中，有 28 个组织的近乎重复发生的恐怖主义事件达到了统计显着水平。尽管我们的研究结果表明，近重复攻击在组织中很常见，但我们发现，近重复攻击中使用的策略（例如，武器选择、目标类型）在组织之间差异很大。这项工作的结果提供了见解，增强了我们对恐怖主义模式的了解，从而改善了反恐行动。此外，我们的端到端数据驱动框架提供了强大的决策支持工具，用户可以首先检测近乎重复的攻击，然后识别这些攻击中的关键操作模式（对于任何感兴趣的恐怖组织）。