Skip to Main Content
The stable model semantics for normal programs has the problem that logical consequences of programs cannot, in general, be stored as lemmas, because the set of stable models of the resulting program may change. We argue that it is possible to assert a conclusion A as a lemma in the stable model semantics, if asserting at the same time a set of facts supporting the conclusion (that we call a base set for A). The effect on the meaning of the program is that of selecting some of the stable models containing A. The collection of all base sets for A generates all the stable models containing A. We propose a characterization of base sets that identifies minimal ones, i.e. the fewest and smallest base sets for A.