Granular Computing as a Basis for Consistent Classification Problems
Yiyu Y. Yao, JingTao Yao · 2002
Within a granular computing model of data mining, we reformulate the consistent classification problems. The granulation structures are partitions of a universe. A solution to a consistent classification problem is a definable partition. Such a solution can be obtained by searching a particular partition lattice. The new formulation enables us to precisely and concisely define many notions, and to present a more general framework for classification.