Information Granules and Approximations in Incomplete Information Systems
Wei-Zhi Wu, Xiaoping Yang · 2007
In this paper three types of information granular structures, called similarity classes, maximal consistent blocks, and labeled blocks, in incomplete information systems are introduced. Their properties are examined. Based on the three structures of granules, three types of rough set approximation models are derived for mining of certain and possible rules in incomplete decision tables. The relationships among the three rough set models are established.