UDC 519.226.3; 519.816
Authors
G.I. Kozhomberdieva D.P. Burakov, G.A. Khamchichev
Abstract
The article discusses the possibility of applying the Bayesian approach to solving the task of rating objects by a group of experts. The result of evaluating each object is an a posteriori probability distribution on the set of hypotheses about the rank of this object. The values of conditional probabilities used in calculating the posterior distribution are understood as the degrees of compliance of the obtained evidence (expert estimates) with the assumptions about the truth of the hypotheses. Each expert has the right, in case of uncertainty, to assign
adjacent ranks, which is an advantage of the approach. In addition, in order to resolve conflicts, it is possible to take into account the unequal competence of experts, and, if necessary, perform multi-round assessment. The final rating of objects is formed as a result of simple processing of all received posterior distributions. An example of using the proposed approach and a description of its software implementation are given.
Keywords
rating, expert evaluation, individual and group preferences, small group decision making, Bayesian approach, Bayes formula.
