Automated Selection of Grounding Algorithm in Answer Set Programming

Answer Set Programming (ASP) is a powerful language for knowledge representation and reasoning. ASP is exploited in real-world applications and is also attracting the interest of industry thanks to the availability of efficient implementations. ASP systems compute solutions relying on two modules: a grounder that produces, by removing variables from the rules, a ground program equivalent to the input one; and a model generator (or solver) that computes the solutions of such propositional program. In this paper we make a first step toward the exploitation of automated selection techniques to the grounding module. We rely on two well-known ASP grounders, namely the grounder of the DLV system and GrinGo and we leverage on automated classification algorithms to devise and implement an automatic procedure for selecting the "best" grounder for each problem instance. An experimental analysis, conducted on benchmarks and solvers from the 3rd ASP Competition, shows that our approach improves the evaluation performance independently from the solver associated with our grounder selector.