Conversion-based aggregation algorithms for linear ordinal rankings combined with granular computing

Abstract

This paper proposes two aggregation algorithms for aggregating individuals’ linear ordinal rankings (LORs) in group decision-making environment, which improves the computability of LORs. At first, in order to represent the position information of alternatives, we depict LORs by extended preference map, and then analyze the basic statistical quantitative characteristics of the LORs. Subsequently, combining with the concept of granular computing, we derive the arithmetic expressions of interval utility values for alternatives, in which the information granularity indices are introduced to determine the interval lengths. Later on, two programming​ models are established with the aim of minimizing the differences between individual opinions and the corresponding collective one. Besides, in the two models, we propose the methods to determine experts’ weights according to the differences of experts’ knowledge backgrounds. Both information granularity index and the aggregated interval utility value can be calculated by solving the models, based on which the final aggregated ranking can be determined. Furthermore, a numerical case concerning the electric vehicle charging station site selection problem is presented to illustrate the usage of the proposed algorithms, and finally, the efficiency and features of the two algorithms are exhibited through comparative analyses and simulation experiment.