Generalized multi-granulation double-quantitative decision-theoretic rough set of multi-source information system

Abstract Traditionally, multi-source information system (MsIS) is typically integrated into a single information table for knowledge acquisition. Therefore, discovering knowledge directly from MsIS without information loss is a valuable research direction. In this paper, we propose the generalized multi-granulation double-quantitative decision-theoretic rough set of multi-source information system (MS-GMDQ-DTRS) to handle this issue. First, we propose a generalized multi-granulation rough set model for MsIS (MS-GMRS) as the basis of other models. In this model, each single information system is treated as a granular structure. Next, we combine MS-GMRS with double-quantitative decision-theoretic rough set to obtain two new models. They have better fault tolerance capability compared with MS-GMRS. Furthermore, we propose corresponding algorithms to calculate the approximation accuracy of the proposed models. Experiments are carried out on four datasets downloaded from UCI. Experimental results show that the two new models have better fault tolerance in directly acquiring knowledge from MsIS.