PAC-learning of preference relations over interpretations in lazy nonmonotonic reasoning
Ken Satoh · 2000
Abstract This paper presents a method of learning a class of preference order over interpretations in a propositional language which is a generalization of circumscription. The order considered in this paper is based on the distance in a weighted Euclidean space between the most preferable interpretation among all the interpretations and an interpretation represented as lattice points whose elements take values of either O or 1. Nonmonotonic reasoning defined by this order is a subclass of lazy nonmonotonic reasoning (Satoh 1992) where nonmonotonicity emerges only when new added knowledge is contradictory to the previous belief.