Temporal abstract argumentation frameworks

Abstract

For decision-making processes to be automated in critical real-world systems, autonomous agents must not only be powerful and efficient, but transparent, accountable, unbiased, explainable, ethical, and capable of handling inconsistent, unreliable and incomplete information. A possibly suitable formalism for modeling reasoning with such features is the Abstract Argumentation Framework, in which arguments without intricate structures are evaluated solely based on their interaction with each other. Justifications for or against claims can be naturally obtained in a format similar to human discourse. The expressiveness of these frameworks has been expanded to incorporate various notions such as support, preferences and weights. The inherent temporal aspect in human reasoning led to the development of temporal frameworks in which arguments’ acceptability and interactions are relative to time. Current approaches of including time in argumentation allow for modeling argument or attack availability, but general temporal interactions between arguments have not been considered in traditional abstract argumentation frameworks. This work reviews the main approaches of featuring time in argumentation from the literature and discusses how a simple translation approach is able to represent variegated temporal interactions.