Affordable classes of normal logic programs

Abstract

Limiting the number of times a variable appears in either the head or the body of a rule, we identify two classes of normal propositional logic programs. These classes have the desirable property that stable models, if they exist, can be found in linear time (worst case). We also identify a related class containing programs for which the well-founded model can be acquired in linear time, yet for which computing the stable model(s) remains NP-complete. We show in this class how, by relaxing one constraint, previously linear complexity is increased to intractability.