LEARNING DOMAIN-SPECIFIC HEURISTICS FOR ANSWER SET SOLVERS

Abstract

In spite of the recent improvements in the performance of Answer Set Programming (ASP) solvers, when the search space is sufficiently large, it is still possible for the search algorithm to mistakenly focus on areas of the search space that contain no solutions or very few. When that happens, performance degrades substantially, even to the point that the solver may need to be ter- minated before returning an answer. This prospect is a concern when one is considering using such a solver in an industrial setting, where users typically expect consistent performance. To overcome this problem, in this paper we propose a technique that allows learning domain-specific heuristics for ASP solvers. The learning is done off-line, on representative instances from the target domain, and the learned heuristics are then used for choice-point selection. In our experiments, the introduction of domain-specific heuristics improved performance on hard instances by up to 3 orders of magnitude (and 2 on average), nearly completely eliminating the cases in which the solver had to be terminated because the wait for an answer had become unacceptable.