Most of collective decision procedures used in group decision making require a full ranking or rating of the alternatives from the agents. However, the complete specification of preferences can be a cognitively difficult task when the number of alternatives increases. It is particularly the case on combinatorial domains, as alternatives are very numerous and implicitly defined. Moreover, the determination of a winner does not necessary requires the agents to express their preferences over the entire set of alternatives. Even when preference information is too poor to determine a winner, it may be sufficient to rule out some alternatives. Additionally, a partial specification of preferences would allow them to keep a certain privacy. This shows the importance of designing computer-aided elicitation procedures aiming to focus the elicitation burden on the useful part of preference information and motivates the topic of my PhD.
展开▼