Methodology for processing expert judgments based on fuzzy weighted summation

Abstract

The paper is devoted to the development and research of a new technique for fuzzy ranking of alternatives. The authors of this approach suggest dividing the criteria space into fuzzy preference areas. A criterion for the proximity of fuzzy preference ratings is also formulated. The scientific novelty of the proposed approach lies in the proposed procedure for clarifying the level of preference based on fuzzy weighted summation, for alternatives with close fuzzy levels of preference. The solution of the problem in this statement makes it possible to increase the distinguishability of alternatives by preference. Two approaches to the construction of membership functions of fuzzy preferences were investigated: the first method - choosing the type of membership function and then entering the parameters of the membership function; the second method is entering the characteristic points of the membership function and constructing a piecewise polynomial approximation. After entering preferences and presetting, this method allows ranking an arbitrary number of alternatives in a mode close to real time. In cases of high dimensionality of the vector criterion and optimized parameters of the domain model, as well as in the presence of complex algorithmic rules, the solution of such problems is of high scientific importance. In the public domain at http://ws-dss.com one can use this method - fuzzy ranking based on preference areas. The proposed approach was implemented (tested) on the problem of selecting an electronic pilot tablet.