摘要
针对经典的数据流挖掘算法Lossy Counting算法空间性能较差,并且在搜索指定长度的频繁项所用时间较长等缺点.提出了基于改进的有向图结构的数据流挖掘算法.改进的算法可在图中双向查询和增加频繁项,并且结构中包含了频繁项的长度,在进行指定长度频繁项查询时,无需遍历整个数据结构.实验表明,改进算法比Lossy Counting算法执行效率有所提高.
In view of the classic data stream mining algorithm Counting Lossy algorithm has a poor perfor-mance in space,and in the search for the specified length of the frequent items with a long timeand other shortcomings.A data stream mining algorithm based on the improved directed graph isproposed.The improved algorithm can be bidirectional query and add frequent items in the graph,and the structure contains the length of the frequent items,which is not required to traverse the en-tire data structure when the frequent item of the specified length is carried out.The experimentsshow that the improved algorithm is more efficient than the Counting Lossy algorithm.
出处
《通化师范学院学报》
2018年第2期65-69,共5页
Journal of Tonghua Normal University
基金
2016年安徽省高校省级自然科学研究重点项目(KJ2016A136)
2017年安徽省高校省级自然科学研究重点项目(KJ2017A759)
安徽省质量工程项目(2014mooc093)
关键词
数据流挖掘
关联规则
频繁项集
data stream mining
association rules
frequent itemsets