Incremental computation of warranted arguments in dynamic defeasible argumentation

Defeasible Logic Programming (DeLP) is an argumentation-based reasoning tool that allows users to contemplate reasons for and against certain conclusions---the warrant status of literals in the knowledge base is the main output of this dialectical process. Since the computation of these warrant statuses is a costly endeavor, any update to the knowledge base has a huge impact if carried out naively. In this paper, we study the case of updates consisting only of additions of either strict or defeasible rules, identifying conditions under which we can avoid wasted effort and proposing a data structure to keep track of which literals' warrant status can potentially be affected by a given update. We report the results of a preliminary set of experiments showing that our incremental algorithm affords significantly lower running times in practice.