Lazy Model Expansion by Incremental Grounding

摘要

Ground-and-solve methods used in state-of-the-art Answer Set Programming and model expansion systems proceed by rewriting the problem specification into a ground format and afterwards applying search. A disadvantage of such approaches is that the rewriting step blows up the original specification for large input domains and is unfeasible in case of infinite domains. In this paper we describe a lazy approach to model expansion in the context of first-order logic that can cope with large and infinite problem domains. The method interleaves grounding and search, incrementally extending the current partial grounding only when necessary. It often allows to solve the original problem without creating the full grounding and is hence more widely applicable than ground-and-solve. We report on an existing implementation within the IDP system and on experiments that show the promise of the method.