A unifying view for logic programming with non-monotonic reasoning (Q1390932)

The paper provides a simple formulation of a unified LP-based semantic framework within which three main extensions of logic programming for non-monotonic reasoning are treated uniformly, namely two forms of negation and abduction. The semantics is based on the well-known notion of Herbrand models for definite Horn programs, through the notion of positive version of a program. The set of Herbrand models for definite (Horn) programs is restricted in a step-wise fashion in order to identify the intended models of the original program. An interesting issue related to a semantic characterization is the study of the equivalence relation that is induced with particular reference to compositionality and fully abstractness properties. This topic is discussed in the paper, in particular, it is shown that the compositionality property can be easily accommodated in the proposed semantic framework, thanks to the model-theoretic grounds. An abstract specification of a proof-procedure for the overall framework is developed. The proof-procedure is shown to be sound and weakly complete with respect to the model-theory.