摘要
为了提高半结构化文档数据流的挖掘效率,对原有挖掘算法StreamT进行了改进,提出了一种半结构化文档数据流的快速频繁模式挖掘算法———FStreamT.该算法针对利用集合存储候选频繁模式效率较低的缺点,采用枚举树存储候选频繁模式,可以有效地提高对候选频繁模式集合进行查找和更新的效率,同时利用频繁模式的单调性和枚举树的特点减小了维护负边界的搜索空间,从而提高了整个算法的效率.理论分析和实验结果表明,算法FStreamT与算法StreamT相比具有较高的效率,是有效可行的.
To improve the efficiency of the semi-structured data stream mining, a fast algorithm for mining frequent patterns from semi-structured data stream, FStreamT, is proposed based on StreamT. To solve the problem of low efficiency of storing frequent patterns in set, this algorithm stores frequent patterns in enumeration tree, which is more efficient when searching and updating the frequent pattern set, and at the same time reduces the search space of maintaining the negative border using the monotonicity of frequent pattern and the characteristics of enumeration tree. Theoretical analysis and experimental results show that the FStreamT algorithm is feasible and more efficient than the StreamT algorithm.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第3期452-456,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(70371015)
关键词
数据挖掘
频繁模式
数据流
枚举树
data mining
frequent pattern
data stream
enumeration tree
