Improved Rough Set Model Based on Set Pair Connection Degree

In order to apply the classical rough set theory in the incomplete information system, equivalence relation has been extended, such as rough set model based on set pair connection degree. But as the binary relation defined in this model has some limitations, inconsistent problems will happen in the process of decision making and when null values are too many the performance is not desirable. In this paper, the reason of these limitations generation is analyzed, and in view of these limitations, a new binary relation based on set pair connection degree, called generalized set pair similarity relation, is proposed. Based on this a more generalized rough set model is presented. Finally, the compare of the generalized rough set mode with some existing extension of rough set models is given. By an example, it is demonstrated that the new model is simpler and more effective when processing incomplete information system.