AOG-ags algorithms and applications

Abstract

The attribute-oriented generalization (AOG for short) method is one of the most important data mining methods. In this paper, a reasonable approach of AOG (AOG-ags, attribute-oriented generalization based on attributes' generalization sequence), which expands the traditional AOG method efficiently, is proposed. By introducing equivalence partition trees, an optimization algorithm of the AOG-ags is devised. Defining interestingness of attributes' generalization sequences, the selection problem of attributes' generalization sequences is solved. Extensive experimental results show that the AOG-ags are useful and efficient. Particularly, by using the AOG-ags algorithm in a plant distributing dataset, some distributing rules for the species of plants in an area are found interesting. © Springer-Verlag Berlin Heidelberg 2007.