Relative possibility

The authors describe a new uncertain knowledge representing method, called RES (relative evidential support, or relative evidence strength), based on representing arguments and their relative strengths. It is shown that in one of its simplest forms, RES represents the same sort of uncertain information as that represented by a possibility distribution, though the starting points of these two models are quite different. RES is compared to the possibility model, showing that while numbers are convenient, they are not always trustworthy and sometime are even undesirable. Using an example, it is also shown that RES may reflect some intuitive meaning of defaults.<>