Supporting decision making in combinatorially explosive multicriteria situations

摘要: Abstract Several real-world problems, including distributed system design and product among others, are characterized by combinatorially explosive solution spaces as well multiple, conflicting criteria. Strategies for finding near-optimal solutions, developed combinatorial not applicable in such situations, which require a balance between extensive computation continual interaction. This makes support or automation of these decisions difficult task. Current approaches to solve problems fall three categories: analytical, genetic algorithm-based local generators. They frequently assume well-behaved functions clear understanding interdependencies Many however, present noisy discontinuous evaluation contain holistic functions, rendering inadequate problems. We propose theoretically grounded approach decision this class The combines broad deep searches with decision-maker feedback that allows the guide and/or stop search. Specifically, it provides information about (a) search explored/probed so far, (b) space yet explored (or may never be explored). operationalize two-phase procedure. first phase — characterization requires choices randomization, sampling estimation techniques. second iterative probes heuristics, fuzzy interpretations based on can evaluate alternatives demonstrate specific instantiation multicriteria object assignment problem verify feasibility our approach.