Computing Answer Sets for Monadic Logic Programs via MapReduce

ABSTRACT. In this paper the applicability of the MapReduce framework for parallel computation for monadic logic programs is studied. In monadic programs all predicates are of arity one. Two different approaches are suggested: the first is based on parallelizing on constants, the second parallelizes on predicates. In each case, a method is provided that makes use of MapReduce and calls standard ASP solvers as back-ends via an abstract API. For each method, we provide a theoretical analysis of its computational impact.