Entropy-based selection with multiple objectives

In this article the authors present an approach to determine the initially unspecified weights in an additive measurable multiattribute value function. They formulate and solve a series of nonlinear programming problems which (1) incorporate whatever partial information concerning the attribute weights or overall relative value of alternatives the decision maker chooses to provide, yet (2) yield a specified set of weights as a result. Although each formulation is rather easily solved using the nonlinear programming software GINO (general interactive optimizer), solutions in closed form dependent on a single parameter are also provided for a number of these problems.