Argumentation-Based Answer Set Justification
Cláudia Schulz · 2012
In the fields of knowledge representation and nonmonotonic reasoning, answer set programming is a widely used technique for solving problems. It captures problems in a declarative way, providing solutions based on the stable model semantics. As an alternative to constructing answer sets by hand, answer set solvers like CLINGO can be used to automatically compute solutions. No matter how an answer set is constructed, it will always be a plain set of literals, which means that there is no explanation why it is the solution for a problem. It is therefore desirable to have a justification for an answer set, especially if it is not the expected solution. This issue is closely related to answer set debugging, where the computation of answer sets is used as a form of justification. In contrast, the aim of my work is to provide a general justification of answer sets, independent of the way they are constructed. Another current research topic in knowledge representation is argumentation theory. Arguments are constructed from rules and facts, where the semantics is given in terms of sets of arguments. One of these semantics is the stable extension, which has its roots in the concept of stable models. My work investigates the correlation between stable extensions and answer sets for both logic programs with and without preferences. Argumentation-Based Answer Set Justification is intro-duced as a novel way for explaining answer sets of logic programs without preferences. It is based on the stable extensions of an ASPIC+ framework derived from a logic program. In addition, JASA is established as an implementation of the justification approach, based on the answer set solver CLINGO and the APSIC+ tool TOAST. With respect to logic programs with preferences, a detailed comparison between different preference-handling semantics is given in my work. This leads to an implementation which is able to compute preferred answer sets in terms of the preference-handling strategy chosen by the user. My work concludes with the suggestion of Argumentation-Based Answer Set Justification as a method for explaining preferred answer sets. ii