Multigranulation Sequential Three-Way Decisions in Interval-Valued System

Abstract

Three-way decisions model adopts the idea of''rule by three divisions'' and "simplify complexity which provides a multi-level and multi granularity thinking framework and implementation method for solving complex uncertain problems. Within this framework, how to extend the application of the generalized model from decision system to interval-valued decision system is an important issue. In this paper, we propose multigranulation sequential three-way decision models in inter-valvalued information system. Firstly, three types of similarity relations from the view of optimistic, pessimistic and weighted are defined. Based on these similarity relations, the lower and upper approximations under multiple granular structures are computed. Then optimistic, pessimistic and weighted arithmetic strategies are adopted to compute three disjoint regions in each level of multigranulation sequential three-way decisions. Furthermore, nine types of multigranulation sequential threeway decision models were analysed. Finally, the experimental results show that the size of the probabilistic regions varies with the similarity relations and aggregation strategies.