基于多层次关联规则挖掘的反恐情报跨层特征关联分析  

Association Analysis of Hierarchical Attributes in the Field of Counter-Terrorism Intelligence Based on Multi-level Association Rule Mining

作　　者：李勇男[1] LI Yong-nan(School of National Security,People's Public Security University of China,Beijing 100038 China)

机构地区：[1]中国人民公安大学国家安全学院,北京100038

出　　处：《情报科学》2021年第11期127-132,共6页Information Science

基　　金：北京市社会科学基金项目“大数据背景下北京反恐风险特征识别及应急防范机制研究”(21GLC064);中国人民公安大学基本科研业务费重大项目“基于群智协同的大数据驱动的’情景-应对’型反恐决策机制研究”(2021JKF106);2021年公安学科基础理论研究专项项目“大数据背景下的反恐风险预警基础理论研究”(2021XKZX07)

摘　　要：【目的/意义】为了发现更全面、更具有普适性的反恐情报信息,本文在单层次关联规则挖掘的基础上研究反恐情报的多层次关联规则挖掘方法。【方法/过程】根据反恐情报的数据特征提出统一最小支持度和多单项最小支持度参数并用的方式筛选多层次涉恐特征频繁项集,在情报分析过程中保存部分特殊的冗余频繁项集、冗余多层次关联规则和无趣多层次关联规则。【结果/结论】本文的研究可以发现涉恐数据中不同概念分层的关联规律。[创新/局限]文中提出的关联分析方法能够弥补普通的单层次关联规则挖掘在分析包含多层属性的涉恐数据中存在的不足,为反恐预警和反恐决策提供更丰富、更科学、覆盖范围更广的参考。【Purpose/significance】In order to find more abundant and applicable counter-terrorism intelligence information,this paper focuses on the mining method of multi-level association rule based on the single-level association rule mining.【Method/process】According to the data characteristics of counter-terrorism intelligence,a method of using uniform minimum support degree and multiple single minimum support degree parameters is proposed to select frequent itemsets of multi-level terror-related features.In the mining process,it should save some special redundant frequent itemsets,redundant multi-level association rules and uninteresting multi-level association rules.【Result/conclusion】This method could find the correlation relationship of different concept hierarchy in the terror-related data.【Innovation/limitation】This method could make up the deficiency of the common single-level association rule mining in the analysis of the terror-related data containing multi-layer attributes,and provide more scientific and abundant comprehensive references for the counter-terrorism early warning and counter-terrorism decision-making.

关 键 词：数据挖掘 反恐情报 多层次关联规则 概念层次树 频繁项集 支持度 

分 类 号：G250.2[文化科学—图书馆学]