Discovering Closed Patterns over Relational Databases with Generalization Hierarchy

Abstract

In this paper, we study enumeration of a class of generalized patterns on relational databases with generalization hierarchy. Such closed patterns are introduced for data anonymization by local and global encoding schema in privacy preserving data publication. However, there has been no polynomial delay and space algorithms for such closed generalized patterns. After reviewing closed item set mining, we fomally define the class of closed generalized patterns, show their basic properties, and then present an efficent depth-first search algorithm that exactly enumerates all closed partitions under monotone constraints.