Outranking-based interactive procedure for multiple objective programs

The paper proposes a new approach to solving optimization problems formulated in terms of multiple objective mathematical programming. It is based on the use of an outranking relation for modelling decision maker’s preferences in early stages of an interactive process. This model of preferences is translated into a cone of scalarizing function weights generating a subset of nondominated points of greatest interest, using an ordinal regression method. The presented methodology applies both to linear and to nonlinear (including discrete) programming problems. It could be classified as a weighting vector space reduction method with a visual constructive interaction.