On the incremental computation of semantics in dynamic argumentation

Abstract

Argumentation frameworks often model dynamic situations where arguments and their relationships (e.g., attacks) frequently change over time. As a consequence, the sets of conclusions (e.g., extensions of abstract argumentation frameworks, or warranted literals for structured argumentation frameworks) often need to be computed again after performing an update. However, as most of the argumentation semantics proposed so far suffer from high computational complexity, computing the set of conclusions from scratch is costly in general. In this work, we address the problems of efficiently recomputing extensions of dynamic abstract argumentation frameworks and warranted literals in dynamic defeasible knowledge bases. In particular, we first present an incremental algorithmic solution whose main idea is that of using an initial extension and the update to identify a (potentially small) portion of an abstract argumentation framework, which is sufficient to compute an extension of the updated framework.