中文

Journal of Applied Sciences ›› 2006, Vol. 24 ›› Issue (5): 498-502.

• Articles • Previous Articles     Next Articles

Online Detection of Data Stream Changes Based on Maximum Frequent Itemset Entropy

LIU Xue-jun1,2, XU Hong-bing1, DONG Yi-sheng1, QIAN Jiang-bo1, WANG Yong-li1   

  1. 1. Department of Computer Science and Technology, Southeast University, Nanjing 210096, China;
    2. College of Information Science and Engineering, Nanjing University of Technology, Nanjing 210009, China
  • Received:2005-06-14 Revised:2005-09-19 Online:2006-09-30 Published:2006-09-30

PDF

111

Cited

2

Abstract: Online detection of data stream changes is a new topic in data stream studies, which provides a salient feature compared to other types of data mining.In this paper, a novel method for detection and estimation of data stream changes is proposed.The main concerns include:1) adoption of a novel discrepancy measure for data streams, 2) a new algorithm which can effectively explore and store all maximum frequent itemsets for data streams, and 3) a method for detection of changes based on maximum frequent itemsets information entropy.No previous work has been reported to the authors' best knowledge using maximum frequent itemsets entropy model in detecting data stream changes.Experiments were carried out to study temporal and spatial efficiency of the algorithm.

Key words: change detection, maximum frequent itemsets, data stream, data stream analysis

CLC Number: