A hybrid approach for linguistic information integration to multi-experts multi-attribute decision-making problem

The decision problem is certainly to deal with multiple dimensions of attribute providing by multiple of sources of information in order to obtain the single solution, so-called multi-experts multi-attribute decision-making. It is quite also that the provided information is usually conflict between not only from multiple dimensions but also from the different sources which cause the uncertain of information. Besides, the most of obtained information is naturally in the linguistic forms that always cause the information loss problem as intensively mentioned in the literature. It is highly necessary to develop the tool in order to deal with linguistic decision making without loss of information. In this study the alternative approach is proposed to overcome such issues by eliminating the linguistic representation and approximation processes of linguistic values in the computational step. To do so, we shall apply the Dempster-Shafer (D-S) theory of evidence as an alternative framework, by first defining experts' preferences on each alternative according to each attribute as randomly linguistic preferences. Then obtaining a collective preference value by making use of Dempster's rule of combination. For decision making purpose, based on pignistic transformation and satisfactory principle, we can provide a rank ordering among the alternatives. A numerical example for tank evaluation problem is used to illuminate the proposed technique.