The Complexity of Model-Preference Default Theories

Most formal theories of default inference have very poor computational properties, and are easily shown to be intractable, or worse, undecidable. We are therefore investigating limited but efficiently computable theories of default reasoning. This paper defines systems of Propositional Model Preference Defaults, which provide a true model-theoretic account of default inference with exceptions. Some of our results extend to other nonmonotonic formalisms, such as Default Logic.