Argumentation-Based Explanations for Answer Sets Using ADF

Abstract

This paper presents so-called asl-explanation graphs for answer set programming based on a translation of extended logic programs to abstract dialectical frameworks (ADF). The graphs show how a literal can be derived from the program, and they evaluate in an argumentative way why necessary assumptions about literals not contained in an answer set hold. With the set of all asl-explanation graphs for a literal and an answer set, it is possible to explain and justify thoroughly why the literal is or is not contained in that answer set. Additionally, we provide a criterion to improve the clarity of explanations by pruning nodes without loss of information and selecting most significant asl-explanation graphs.