基于位置信息的显露序列模式挖掘研究

摘要

显露序列因为具有强区分能力,常被用来构建有效的分类器.当前算法大多关注序列模式的支持度或出现次数,而忽略序列模式在序列中的出现位置,这将导致一些重要的信息丢失.为此,提出一种带有局部位置信息的显露序列模式,并给出位置显露序列模式挖掘算法.该算法基于出现次数框架,结合后缀树,省略了候选模式的生成与选择步骤,能够快速有效地挖掘出位置显露序列模式.实验结果表明,采用位置显露序列模式构建的分类器在平均分类准确度上高于传统的显露序列模式挖掘算法.%Owing to the strong ability of distinguishing,emerging patterns have been widely used to build defective classifier.As most of the existing algorithms focus on the support or the occurrences of sequence patterns,and the location of the sequence patterns in a sequence is usually ignored,some important information may be missed.In this paper,we put forward an emerging sequence pattern with local location information,and a mining algorithm of the emerging sequence pattern with location information.Based on the framework of occurrences,combined with the suffix tree,omitting the generation and selection procedure of candidate patterns,this algorithm can quickly and efficiently mine emerging sequence patterns with the location information.The experimental results show that the classifier which is built by emerging sequence patterns with location information is better than the traditional algorithm of mining the emerging sequence patterns on the average classification accuracy.