10 March 2006, Computational Social Choice Seminar, Ulle Endriss
Social welfare orderings are a means of formalising what should be considered a "good" agreement from a social point of view, given the preferences of the individual members of society. In an egalitarian system, for instance, we would strive for an agreement that maximises the utility of the weakest member of society, while in a utilitarian system we would want to maximise average utility. While the study of social welfare orderings originated in Social Choice Theory and Welfare Economics, they have recently also found important applications in Artificial Intelligence (and in Multiagent Systems in particular). Furthermore, the axiomatic approach to characterising social welfare orderings should be of great interest to logicians. In this talk, I shall introduce the concepts of social welfare ordering and collective utility function, and explain some of the fundamental results in the field pertaining to representability and axiomatisation issues.
(This talk will be similar to the second lecture in my Multiagent Systems course.)