Nonlinear information aggregation via evidential reasoning in multiattribute decision analysis under uncertainty

In many decision situations, it is inevitable to deal with both quantitative and qualitative information under uncertainty. Evidence-based reasoning within a multiple criteria decision analysis framework provides an alternative way of handling such information systematically and consistently. In this paper, the evidential reasoning (ER) approach is introduced, which is based on a recursive ER algorithm that, in essence, constitutes a nonlinear information aggregation process. To facilitate the application of the ER approach and as an indispensable part of its development, the nonlinear features of the ER information aggregation process need to be thoroughly investigated and properly understood. This forms the theme of this paper where the nonlinear features are explored by examining typical reasoning patterns in aggregating harmonic, quasi-harmonic, and contradictory decision information. This analytical investigation provides insights into the recursive nature of the ER approach as well as valuable experience that could be useful to other researchers and practitioners interested in developing and applying operation research/artificial intelligence (OR/AI)-based approaches for decision analysis under uncertainty. The analytical study is complemented by the numerical studies of two application examples. The analysis of a quality assessment problem for motor engines is aimed to show the step-by-step process of implementing the ER approach and to illustrate its nonlinear features in a real-life decision situation. The study of a more complex assessment problem in ship design is intended to demonstrate the potential of the ER approach and its supporting software for dealing with general decision problems.