On Graphical Modeling of Preference and Comfort

Abstract

User's choices involve habitual behavior and genuine decisions. Habitual behavior is often represented using preferences. The conditional preference network (CP-net) is a graphical tool for representing and reasoning about conditional preferences. We argue that genuine decisions are environmentally influenced and we introduce the notion of "comfort" to represent this type of choices. In this paper, we propose a graphical model that we call CP-net with Comfort (CPC-net), to represent and reason about both user's preferences and comfort. The CPC-net extends the CP-net structure with a semantics that accounts user's comfort together with preferences. We indeed show that the CP-net is a special instance of the CPC-net. A CPC-net consists of two optimal outcomes, one for preference and another for comfort. These outcomes can be obtained in time linear on the number of variables for an acyclic CPC-net.