摘要：A learning-oriented interactive method is proposed intended to solve discrete multicriteria choice problems with a large number of discrete alternatives and a few quantitative criteria. The method suggested is inspired by the partition-based methods designed to solve multiple objective mathematical programming problems. At each iteration, the DM may choose the current preferred alternative from one ranked set or from two ranked sets of alternatives. The first ranked set of alternatives is obtained by solving a discrete optimization scalarizing problem based on the preference information given by the DM about the desired changes of the values, desired directions of changes and desired intervals of changes for some or for all of the criteria in relation to their values in the current preferred alternative. The second ranked set is obtained using AHP or an outranking procedure when the DM is willing (able) to provide additional preference information, e.g., pairwise comparisons of the criteria or inter- and intra-criteria information. The DM can successively and systematically screen the set of non-dominated alternatives by the proposed method. The method is illustrated with the help of an example.
