Local Weighted Generalized Multigranulation Neighborhood Rough Set

Local neighborhood rough set is an emerging rough set model, which provides more efficient knowledge representation for weakly labeled data with real-valued attributes in big data. It mainly studies knowledge representation under single granulation, but it cannot effectively and comprehensively represent knowledge of multiple granulations. In this paper, having considered the importance of different granulations for knowledge representation and decisions, we propose a Local Weighted Generalized Multigranulation Neighborhood Rough Set (LWGMNRS) model by using a novel weighted granulation method for knowledge representation of weakly labeled data with multiple granulations. Meanwhile, we discuss the relationships among LWGMNRS and the existing rough set models in detail, as well as its advantages and significances. Finally, theoretical analysis and case study show that the sig-nificance of LWGMNRS in knowledge representation.