نمونه متن انگلیسی مقاله
Multiple Criteria Decision Analysis methods, such as ELECTRE, PROMETEE, AHP, TOPSIS, VIKOR, have been applied to solving numerous real-life decision making problems in business and management. However, the mechanics of those methods is not easily understandable and it is often seen by users without much formal training as a kind of “scientific witchcraft”. In order to make those popular MCDA methods more transparent, we provide a simple framework for interpretations of rankings they produce. The framework builds on the classical results of MCDA, in particular on the preference capture mechanism proposed by Zionts and Wallenius in seventies of the last century, based on Simple Additive Weighting. The essence and the potential impact of our contribution is that given a ranking produced by a MCDM method, we show how to derive weights for the Simple Additive Weighting which yield the same ranking as the given method. In that way we establish a common framework for almost no–cost posterior analysis, interpretation and comparison of rankings produced by MCDA methods in the expert systems environment. We show the working of the concept taking the TOPSIS method in focus, but it applies in the same way to any other MCDM method. We illustrate our reasoning with numerical examples taken from literature. 1
2 Multiple Criteria Decision Analysis (MCDA) is a topic well repre3 sented in the field of expert systems; many of them employ MCDA 4 for solving complex problems of decision making (see Alemi5 Ardakani, Milani, Yannacopoulos, & Shokouhi, 2016; Mardani, Ju6 soh, & Zavadskas, 2015; Östermark & Salmela, 1988; Ozernoy, 7 1988). 8 Solving an MCDA problem is usually understood as determining 9 an alternative (a decision variant) which corresponds to the best, 10 in the decision maker’s opinion, combination of (at least two) cri11 teria values, or in a broader sense, as ranking alternatives from the 12 best (in the above meaning) to the worst one. Because attaining 13 the maximal values with respect to all criteria simultaneously, in 14 general, impossible, solving an MCDA problem requires that some 15 information on preferred combinations of criteria values (so called 16 preference information) has to be articulated by the decision maker 17 (DM).