Attribute generalization and fuzziness in data mining contexts

Abstract

This paper shows problems with combination of rule induction and attribute-oriented generalization, where if the given hierarchy includes inconsistencies, then application of hierarchical knowledge generates inconsistent rules. Then, we introduce two approaches to solve this problem, one process of which suggests that combination of rule induction and attribute-oriented generalization can be used to validate concept hiearchy. Interestingly, fuzzy linguistic variables play an important role in solving these problems. © Springer-Verlag Berlin Heidelberg 2007.