Intelligent systems for decision support.
详细信息
文摘
This research is focused on multi-criteria decision-making MCDM) under uncertainties, especially linguistic uncertainties. This problem is very important because many times linguistic information, in addition to numerical information, is an essential input of decision-making. Linguistic information is usually uncertain, and it is necessary to incorporate and propagate this uncertainty during the decision-making process because uncertainty means risk. MCDM problems can be classified into two categories: 1) multi-attribute decision-making MADM), which selects the best alternatives) from a group of candidates using multiple criteria, and 2) multi-objective decision-making MODM), which optimizes conflicting objective functions under constraints. Perceptual Computer, an architecture for computing with words, is implemented in this dissertation for both categories. For MADM, we consider the most general case that the weights for and the inputs to the criteria are a mixture of numbers, intervals, type-1 fuzzy sets and/or words modeled by interval type-2 fuzzy sets. Novel weighted averages are proposed to aggregate this diverse and uncertain information so that the overall performance of each alternative can be computed and ranked. For MODM, we consider how to represent the dynamics of a process objective function) by IF-THEN rules and then how to perform reasoning based on these rules, i.e., to compute the objective function for new linguistic inputs. Two approaches for extracting IF-THEN rules are proposed: 1) linguistic summarization to extract rules from data, and 2) knowledge mining to extract rules through survey. Applications are shown for all techniques proposed in this dissertation.