Multigranulation rough set model based on granulation of attributes and granulation of attribute values

Abstract Existing multigranulation rough set models often suppose the family of equivalence relations is existed and do not give the methods to construct the multigranulation space. That is, the multigranulation space is given. So in multigranulation rough set models, how to derive the family of equivalence relations and construct the multigranulation space is an important problem to be resolved. In this paper, considering there are different granulation of attributes and attribute values, I propose multigranulation rough set model based on granulation of attributes and granulation of attribute values (MRAAV), which provides a method to construct the multigranulation space in multigranulation rough set models. Based on MRAAV, multigranulation problem solving space is constructed. The search strategy for problem solving in multigranulation problem solving space is presented, which make the information processing process of rough set is structured and hierarchical. Finally, the rule extraction algorithm based on MRAAV is given, which can not only extract rules from multiple abstract concept levels, but also with low costs. The example and experimental results demonstrate the effectiveness of the proposed model and algorithms.