面向装备作战仿真数据流的改进型频繁项集搜索方法研究  

Study of frequently item set search method for equipment combat simulation

作　　者：曹波伟[1] 薛青[1] 

机构地区：[1]装甲兵工程学院仿真中心,北京100072

出　　处：《计算机工程与科学》2013年第6期152-155,共4页Computer Engineering & Science

摘　　要：装备作战仿真数据流中的事务性数据属性之间存在着各种各样的关联,数据流关联规则分析的目的是找出这些隐藏的关联关系。针对装备作战仿真数据流关联规则分析时数据流的大数据量、变长度的特点,就关联规则获取过程中如何得到频繁项集这一问题进行了研究。为了在有限空间内快速地从装备作战仿真数据流事务数据中获取频繁项集,针对经典的频繁项集算法Lossy Counting空间占用过大的缺陷,提出了一种基于下三角矩阵的Lossy Counting数据流关联规则频繁项集搜索算法FIBM。该算法通过下三角项集存储结构代替trie树结构,能在较少的空间占用下扫描数据流一次,在线实时分析装备作战仿真数据流,具有较强的实用性。数值实验证明,进行频繁项集搜索时,FIBM算法比Lossy Couting算法的执行效率有了明显的改善。Combat simulation data stream exists a variety of associated between data attributes, so a- nalysis of data stream association rules is designed to find out the hidden relationship. In view of the combat simulation data stream characteristics of large volume data, variable length, this paper studies how to get the frequent itemsets in the the association rules,in order to obtain frequent itemsets from combat simulation data stream within the limited space quickly, in view of space usage shortages of asso clarion rules algorithms Lossy Counting , we proposed a kind of the improved frequent itemsets search method （FIBM）based on the matrix, the method use triangular itemsets matrix instead of the trie tree structure in Lossy Couting algorithm, thereby effectively reducing the Lossy Couting algorithm space occupancy rate. Numerical experiments prove that FIBM algorithm for frequent itemsets search are more improved performance obviously than Lossy Couting algorithm.

关 键 词：装备作战仿真数据流 关联规则 频繁项集搜索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]