In this paper, we present a composite framework of sequential three-way decisions to deal with hybrid data based on the fusion of different granularities. According to the top-down manner, we construct a multilevel composite granular structure by the addition of a new attribute type, and define a general composite binary relation based on three kinds of fusion strategies. At each level, the particular regions including seven selections are considered to induce the acceptance, non-commitment, and rejection rules. Some uncertain objects may be further investigated by more types of attributes at the next level. In this way, such multilevel processing of hybrid data naturally leads to the composite sequential three-way decisions.
展开▼