Efficient Discovery of Closed Hyperclique Patterns in Multidimensional Structured Databases

Structured data is becoming increasingly abundant in many application domains recently. Furthermore, more complex but valuable databases will be obtained by combining plural structured databases. In this paper, we focus on "Multidimensional Structured Databases'' as one of the typical examples of such complex databases, and propose a new data mining problem of finding closed hyperclique patterns, i.e., closed sets of correlated patterns, in them. To solve this problem efficiently, an algorithm named CHPMS is proposed which effectively utilizes the generality ordering and the properties of correlation and closedness. The effectiveness of the proposed algorithm is confirmed through the experiments with real world datasets.