刑事案件的多层关联分析模型研究

Research on Model of Multi-level Association Rules for Criminal Cases

为探寻当前刑事案件的发案规律与特点,以便及时预防和打击刑事犯罪,本文研究提出了刑事案件的多层关联分析模型。首先提出了刑事案件的多层关联规则挖掘的模型框架,依据所建立的刑事案件多维多层数据立方体模型,设计了层间递减支持度策略。基于经典的Apriori算法,提出了适于多层频繁谓词集搜索的改进Apriori算法,按照最小支持度与最小置信度的要求产生强关联规则。由于多层挖掘产生的规则可能存在＂祖孙＂关系,本文设计了结果分析中的减少冗余规则。最后,利用大连公安局提供的甘井子区1999～2006年的18 629条刑事案件的真实数据,验证了模型的正确性与有效性。

In order to find the rules and the characteristics of criminal cases to prevent and combat crime,this paper has built a model of multi-level association rules for criminal cases.First,the framework of multi-level association rules mining is proposed.The strategy of level descending support is designed according to the characteristics of multidimensional and multilevel data model.Then the improved Apriori algorithm is introduced for the search of multilevel frequent predicate sets, and strong association rules are generated according to the requirement of minimum support and minimum confidence.Because of＂grandfather-child＂relationship in multidimensional mining,the checking of redundant rules is necessary in the result analysis.Finally,the correctness and usability of the model is validated on the data of 18629 criminal cases from 1999 to 2006 of Ganjingzi district offered by Dalian Police Bureau.