Knowledge Granulation Based Roughness Measure for Neighborhood Rough Sets

摘要

Neighborhood rough sets have been applied to feature selection and attribute reduction successfully. Roughness is an important uncertainty measure for a concept in an information system. In this paper, generalized from the classical roughness, a new uncertainty measure based on granulation of knowledge for neighborhood rough sets is proposed to overcome the limitations, and then present its properties. Theoretical studies and examples show that the new uncertainty measure is more precise than existing ones.