Rule-Based Attribute-Oriented Induction for Knowledge Discovery

Abstract

This paper introduces a rule-based Attribute-Oriented (AO) Induction method on rule-based concept hierarchies that can be constructed from generalization rules. Based on analyzing some major previous approaches such as rule-based AO induction with backtracking, path-id based AO induction and a cyclic graph based AO induction, we propose a new approach to facilitate induction on the rule based case that can avoid a problem of anomaly and overcome disadvantages of these above methods. Experimental studies show that the new approach is efficient and suitable for providing condensed and qualified summarizations.