系列案件犯罪地理目标模型下空间点群综合距离算法优化 Criminal Geographic Target Model Optimal Integrated Distance between Spatial Points of Serial Burglaries

Abstract

本文根据地理画像理论中的犯罪地理目标模型(CGT)，以国内Q、S两个城市城区的系列入室盗窃案件为研究对象，进行犯罪人的犯罪活动距离以及案件点群综合距离算法优化选择研究，旨在利用系列案件点群的空间分布预测犯罪人可能的居住区域，为警务部门刑侦破案服务。通过研究发现：1) 研究区内系列入室盗窃案件的犯罪人在进行犯罪时遵循距离衰减原则，即在一定的距离范围内集中多次作案；2) 在进行CGT建模时，两研究区模型的经验性常数表现出一致性；3) 在选择最优的案件点群综合距离算法模拟犯罪人“心理缓冲区”问题上，与传统CGT模型使用点群内平均最邻近距离的一半相比，研究区Q市使用标准距离算法获得了更好的预测效果。 We used criminal geographic targeting (CGT) model—one of the geographic profiling methods— based on the real data of serial burglaries of two Chinese cities Q and S to discover the journey- to-crime distance of offenders and the optimal integrated distance between points for CGT model. We wanted to detect offenders’ residences and serve for the police department. Our results show that 1) the serial burglary offenders of our two research areas comply with the distance-decay function when they commit crimes, that is, they always repeatedly commit crimes within a certain distance; 2) the empirical constants of CGT model in the two research areas exhibit consistency; 3) compare with the traditional CGT model, city Q uses standard distance of crime points to simulate the “buffer zone” of offenders can improve the accuracy of CGT model.