基于Spark/GraphX图聚类算法的入室盗窃串并案研究

RESEARCH OF BUNCHING AND MERGING BURGLARY CASE BASED ON SPARK/GRAPHX GRAPH CLUSTERING ALGORITHM

随着我国城镇化进程的不断加速,广泛的人口流动使社会治安环境日趋复杂,犯罪分子系列性作案居高不下,给人民的生命财产安全构成极大的威胁。针对刑事犯罪活动中日益突出的系列入室盗窃案件,提出采用图聚类算法来进行串并案分析。首先利用Spark/Graph X分布式图计算框架,通过提取入室盗窃案的案件特征,计算两两案件之间的相似度,构建案件相似度矩阵;然后依据图论理论,采用图聚类算法实现串并案分析模型。实战工作表明该模型可为侦破案件提供有效的串并线索,极大地减少人工作业,提高了侦查工作的效率。

With the acceleration of the urbanization process in our country,the extensive population flow makes the public security environment become more and more complex,and the serial crimes of criminals are still high,which poses a great threat to the people＇s lives and property safety. In this paper,in view of the increasingly prominent series of burglaries in criminal activities,a graph clustering algorithm is proposed to perform the parallel case analysis. First of all,we used the Spark/Graph X distributed computing framework to extract the case characteristics of burglaries,calculated the similarity between cases,and built the case similarity matrix. Then,according to the graph theory,the graph clustering algorithm was used to implement the parallel case analysis model. The actual combat work shows that the model can provide effective string and clue for detecting cases,greatly reduce manual operation and improve the efficiency of the investigation.