一种高效的挖掘SJEP的算法  

An Efficient Algorithm for Mining SJEP

作　　者：李曼[1] 王大治[2] 

机构地区：[1]中国人民大学信息学院,北京100872 [2]中国科学院成都计算机应用研究所,成都610041

出　　处：《计算机工程》2004年第18期69-70,73,共3页Computer Engineering

摘　　要：SJEP(即最有效的跳跃显露模式)是近年来提出的一种特殊的JEP(即跳跃显露模式),实验表明,仅使用SJEP足以用来建立精确的分类法。因为将现有挖掘JEP的算法直接应用于SJEP无法取得满意的效果,所以该文先压缩存储数据到一棵扩展树结构中,然后通过引入后缀子树的方法,给出了一个高效的SJEP挖掘算法。实验证明,该算法可以在很短的时间内(通常为若干秒)完成学习阶段。SJEP(most Significant Jumping Emerging Pattern), a special JEP(Jumping Emerging Pattern), is proposed recently. Experiment shows tha classifier based on SJEP can achieve higher accuracy. It is not very satisfactory to apply mining JEP algorithms to mine SJEP. In this paper, an efficien algorithm is proposed, which restores data into an extended tree structure and uses postfix subtree. The experiment shows that this algorithm can finis learning phrase in very short time(usually in seconds).

关 键 词：分类 数据挖掘 最有效的跳跃显露模式 

分 类 号：TP311.131[自动化与计算机技术—计算机软件与理论]