The paper belongs to the “Collective choice” section of the mathematical theory of decision-making. A method for non-contradictory aggregation of expert preferences given by quasi-order relations is proposed. The aggregated relation is built according to the rule of “the ma jority of experts”, which satisfies the condition of the minimum distance from expert preferences. Let the expert preferences profile be given by the quasi-order relations ρ1,ρ2, . . .,ρm on the set of alternatives A = {a1,a2,. . . ,an}. Relations will be given by the preference matrices P1, P2, ... , Pm and the vertex adjacency matrices R1, R2,. . . ,Rm of the corresponding digraphs. The preference matrix, in contrast to the adjacency matrix, contains elements 1/2 for equivalent alternatives. The algorithm for constructing a non-contradictory aggregate relation contains the following steps. 1. Construction of a weighted majority digraph G = (A, p∑). The adjacency matrix Rς = ∖r∑∖ of the majority digraph is constructed on the basis of the matrix of total m preferences PΣ = Pk according to the majority rule k=1 rij = 10,, " Pij Pji, where pij are the elements of the matrix P∑. if pij < pji, m The weights on the digraph arcs lij = (pikj -pjki) characterize the degree of superior- k=1ij ji ity of alternative ai over aj∙ and are used to destroy contradictory cycles (Pk = ∖∖pkj ∣∣; 1. j = 1, . . . , n). 2. The destruction of contradictory cycles. Arcs are removed from the cycles that have a minimum weight (with minimal advantage in expert preferences) and belong to the asymmetric part of the relation. In this case, arcs connecting equivalent alternatives are saved. We get a digraph G' = (A, ρ}, R is the adjacency matrix. 3. Construction of aggregated quasi-order p. The adjacency matrix of an aggregate quasi-order is found by the formula R = E V Tr R, where Tr R is the adjacency matrix of the transitive closure of the relation ρ without contradictory cycles. The propositions about the uniqueness and non-contradictory of the constructed aggregated relation p are proved. The computational complexity of the algorithm is O(n3). Based on the constructed aggregated quasi-order relation, the ranking of alternatives is carried out. For this purpose, an algorithm for constructing digraph preference levels has been developed. The algorithm is based on the Demukron procedure of partitioning a digraph without contours into levels N0, N1, . . . , Nt, where k-1 k-1 No = {ai : ai ∈ A, Γai = 0}; Nk = {ai : ai ∈ ANj, Γai Nj}, k = 1,...,t. j=0 j=0 The propositions that allow to modify the Demukron procedure for partitioning into preference levels of an arbitrary digraph are proved. In this case, the condition that equivalent alternatives belong to the same level of preference is satisfied. Using this algorithm, it is possible in particular to build a nonstrict ranking of alternatives. The developed technique can be used to solve multi-criteria problems in case of verbal information about pairwise comparison of alternatives according to the quality criteria.
Download file
Counter downloads: 231
- Title Non-contradictory aggregation of quasi-order relations
- Headline Non-contradictory aggregation of quasi-order relations
- Publesher
Tomsk State University
- Issue Prikladnaya Diskretnaya Matematika - Applied Discrete Mathematics 45
- Date:
- DOI 10.17223/20710410/45/13
Keywords
групповой выбор, мажоритарный граф, агрегированное отношение, квазипорядок, минимальное расстояние, противоречивый контур, уровни предпочтения, col lective choice, weighted majority graph, aggregated relation, quasiorder, contradictory cycle, preference levelsAuthors
References
Ларичев О. И. Вербальный анализ решений / под ред. А. Б. Петровского. М.: Наука, 2006. 400 с
Петровский А.Б. Теория принятия решений. М.: Академия, 2009. 400 с
Миркин Б.Г. Проблема группового выбора. М.: Наука, 1974. 256 с
Мулен Э. Кооперативное принятие решений: Аксиомы и модели. М.: Мир, 1991. 464 с
Нефедов В. Н., Осипова В. А., Смерчинская C. О., Яшина Н. П. Непротиворечивое агрегирование отношений строгого порядка // Изв. вузов. Математика. 2018. № 5. С. 71-85
Нефедов В. Н., Смерчинская C. О., Яшина Н. П. Построение агрегированного отношения, минимально удаленного от экспертных предпочтений // Прикладная дискретная математика. 2018. № 42. С. 120-132
Шрейдер Ю. А. Равенство, сходство, порядок. М.: Наука, 1971. 256 с
Smerchinskaya S.O. and Yashina N. P. An algorithm for pairwise comparison of alternatives in multi-criteria problems // Intern. J. Modeling, Simulation, and Scientific Computing. 2018. V. 9. No. 1. https://www.worldscientific.com/doi/abs/10.1142/S179396231850006X
Нефедов В. Н., Осипова В. А. Курс дискретной математики. М.: Изд-во МАИ, 1992. 262 с

Non-contradictory aggregation of quasi-order relations | Prikladnaya Diskretnaya Matematika - Applied Discrete Mathematics. 2019. № 45. DOI: 10.17223/20710410/45/13
Download full-text version
Counter downloads: 385