On obtaining minimal variability OWA operator weights. (Q1810606)

Ordered weighted average operators are tools used for aggregation in multicriteria decision making. Essentially, they consist of a vector of weights used for aggregation; the weights are assigned to a position in the ordered sequence of aggregates. The orness of an OWA operator is a measure of the degree to which the operator is like an \textit{or} operator. In this paper a class of OWA operators with minimal variance among the weights for a given level of orness is investigated. The problem of finding such weights is formulated as a nonlinear optimization problem. Using Karush-Kuhn-Tucker optimality conditions the authors solve this problem analytically. Thus they determine the exact minimal variability OWA weights for any level of orness.