DATA MINING ON SET- VALUED INFORMATION SYSTEMS

Abstract

Based on the idea of minimizing the original data set, in this paper, we propose a method of selecting representative object set from initial object set to the solve optimal set of attributes problem in set-valued information systems. We demonstrate that the optimal set of attributes on the original objects and the optimal set of attributes on the representative one are equivalent, therefore we confirm the correctness of the method. Because the representative sample size is smaller than the original object’s size, the execution time of algorithms for finding the optimal attribute set on the representative sample is significantly reduced. Representative sample size is large or small depending on the specificity of each real-time information system. At the same time, the article presents the method of exploring ordinal law by converting ordinal monopole information system into binary monopole information system and applying the law biotechnology technique in the systematic set theory based on the binary monotherapy obtained.