Modeling the preferences of agents over a set of alternatives is a principal
concern in many areas. The dominant approach has been to find a single
reward/utility function with the property that alternatives yielding higher
rewards are preferred over alternatives yielding lower rewards