A Prototyping Framework for Reduct-Based ELP Solvers: Methodology and Implementation

Epistemic Logic Programs (ELPs) extend Answer Set Programming (ASP) by incorporating epistemic operators, notably the knowledge operator K. The semantics of ELPs are defined through the concept of world views, which are sets that, in turn, comprise sets of atoms. Various semantic frameworks have been proposed, many of which, including influential early approaches, are based on reduct-based definitions. These approaches generalize the ASP methodology to ELPs by selecting a candidate world view, constructing the corresponding reduct of the program based on this candidate, computing the stable models of this reduct, and subsequently verifying if the initial candidate is indeed a world view. While specialized inference engines (ELP solvers) have been developed for certain semantic approaches, there remains no consensus regarding the 'correct' semantics for ELPs, and new or variant semantics continue to emerge. In response to this evolving situation, this paper introduces a novel fast prototyping methodology that enables the implementation of solvers for any reduct-based semantics. The main advantage of this approach is the ability to rapidly experiment with new semantics on small- to medium-sized programs, as opposed to the very limited small-scale experimentation seen in the literature. This facilitates a thorough preliminary evaluation prior to committing to the more resource-intensive development of dedicated solvers. As a concrete demonstration, we apply our methodology to seminal semantic frameworks already established in the literature. Nevertheless, our approach is readily adaptable to accommodate other reduct-based semantics.