Heuristic algorithms for credulous and sceptical reasoning problems in abstract argumentation

Abstract We consider problems of credulous and sceptical reasoning in abstract argumentation under a variety of semantics and present algorithms for heuristically solving these, i.e. we present algorithms that do not necessarily always give the correct answer but are more performant than correct algorithms. Our algorithms are based on using grounded semantics as a proxy for deciding acceptability w.r.t. other semantics and on bounded search for defenders. We perform a comprehensive experimental evaluation that shows competitive performance of our approaches.