Discovering interesting itemsets based on change in regularity of occurrence

摘要

Mining interesting itemsets/patterns is presented and utilized in a wide range of applications. Organizations and businesses have applied this to observe/track/monitor significant occurrence behavior of objects or events. Currently, with the emergence of new technologies, people may change their needs/behaviors in daily life. Thus, analysis of change on occurrence behavior of objects (or events) can be an important issue in several domains. In this paper, we propose to mine interesting itemsets based on change in regularity of occurrence (called ICROs) to capture change on behavior from actions performed by people. A single-pass algorithm, called MICRO, and a tree structure named ICRO-tree are designed to efficiently mine ICROs. Moreover, a pruning strategy is devised to cut-down search space, computation time and memory consumption. Experiments were done to investigate the performance of MICRO and to show efficiency of MICRO on runtime, memory usage and the number of discovered ICROs.