Decision support system for multicriterial R&D and information systems projects selection

Decision support system described here makes it possible to carry out multicriterial selection of hundreds of projects simultaneously, with tens of criterion functions with bivalent variables including polynomial ones and quotients of linear functions, and tens of resources limitations. Stewart's idea of the special scalarizing function based on the modified reference point approach, which has been hitherto applied for linear benefits and for balance criteria functions only, and its optimization by effective gradient method is used here in the situation described by Santhanam and Kyparisis, involving synergistic effects of second- and third-orders in benefit and cost criterion function, and in resource requirements respecting resource sharing and hierarchical contingency relationships among candidate projects. In addition, the system enables us making the dialogue of the solution in a way of adaptive creation of weights of criterion functions and also flexible projects portfolio changes. A test of efficiency is presented.