期刊名称：Eastern-European Journal of Enterprise Technologies
印刷版ISSN：1729-3774
电子版ISSN：1729-4061
出版年度：2018
卷号：4
期号：4
页码：27-35
DOI：10.15587/1729-4061.2018.140686
语种：English
出版社：PC Technology Center
摘要：We investigated the possibility of fast computation of collective expert estimates of the median type. Despite the widespread use of the Kemeny-Snell and Cook-Seiford medians for calculating the collective expert estimates, the possibilities to reduce the time required to calculate median consensus ranking through the application of the assignment problem and known algorithms for solving it, have been insufficiently investigated. In contrast to the most known methods, the method proposed in this paper is not approximated and retains the original median axiomatics of Kemeny. We investigated a possibility for calculating the Kemeny-Snell and Cook-Seiford medians using the assignment problem applying methods of computer experiment. We estimated the time required to calculate median rankings by four different algorithms for solving the assignment problem. It was established that the proposed method, at a moderate number of alternatives (n<50), computes median rankings over the time close to a real time mode. It is also shown that, in contrast to other methods, the calculation of median rankings using the assignment problem does not depend on coherence degree of individual expert rankings. The results obtained are useful for the practical application of the examined procedure in network expert systems. In such systems, the computation time of a consensus ranking should be close to real time. In addition, in the network expertise systems, due to the random gathering of a team of experts, there is a possibility of the low level of coherence in individual rankings. For the examined procedure, that does not affect the duration of computation. This allows us to recommend the developed computational procedure for a fast search for median consensus rankings by Kemeny-Snell and Cook-Seiford for practical application in the systems of network collective expertise.
关键词：collective expert estimation;median consensus rankings;assignment problem
