The 10th International Conference on Modeling Decisions for Artificial Intelligence Modelització de Decisions per a la Intel·ligència Artificial Barcelona, Catalunya Novembre 20  22, 2013 http://www.mdai.cat/mdai2013 
Termini de submissió:
Pel CDROM (extended): 13 juliol, 2013 
Abstract: To be announced
Authors: Salvador Barberà, Dolors Berga, Bernardo Moreno
Abstract: Restricting the domains of definition of social choice functions is a classical method to test the robustness of impossibility results and to find conditions under which attractive methods to reach collective decisions can be identified, satisfying different sets of desirable properties. We survey a number of domains that we have recently explored, and exhibit the possibility results that emerge for functions defined on each one of them. In particular, we have identified conditions under which the social preference relations derived by different supermajority voting systems would satisfy quasitransitivity, others where individual and group strategyproofness would become equivalent, and still others where the strategyproofness of social choice functions is guaranteed as soon as they satisfy very simple monotonicity and invariance requirements. Our main message is that every specific social choice problem deserves a careful analysis of the domains on which we need to define the method to be used, since this may open the door to attractive possibility results.
Abstract: To be announced
Abstract: To be announced
Abstract: In multiexpert decision making problems, the use of intervalvalued fuzzy sets poses the problem of the nonexistence of a natural total order to rank the alternatives. We consider here the class of admissible orders, which are total orders between intervals defined in terms of two aggregation functions. We make use of such orders to define intervalvalued OWAs and Choquet integrals in such a way that, if the intervals degenerate into single points, we recover the classical concept of such aggregations. We then show that the choice of the linear order determines the solution of the given mutiexpert decision making problem, in a such a way that we can pick any solution up beforehand just by selecting the appropriate linear order. For this reason, we end proposing two algorithms such that second one allows us, by means of the Shapley value, to pick up the best alternative from a set of winning alternatives provided by the first algorithm.
IIIA  Institut d'Investigació en Intel·ligència Artificial
