Towards a Possibilistic Logic Handling of Preferences

摘要: The classical way of encoding preferences in decision theory is by means utility or value functions. However agents are not always able to deliver such a function directly. In this paper, we relate three different ways specifying preferences, namely set particular types constraints on the function, an ordered prioritized goals expressed logical propositions, and subsets possible choices reaching same level satisfaction. These expression modes can be handled weighted setting, here one possibilistic logic. aggregation pertaining criteria then fusing sets goals. Apart from better expressivity, benefits representation put them suitable format for reasoning purposes, modifying them.