Dynamic Argumentation and Action Languages: Towards Explanations

Yann Munro (1) , Camilo Sarmiento (2) , Isabelle Bloch (1) , Gauvain Bourgne (2) , Marie-Jeanne Lesot (1)

1 LFI - Learning, Fuzzy and Intelligent systems
2 ACASA - Agents Cognitifs et Apprentissage Symbolique Automatique

In the context of abstract argumentation, we present the benefits of considering temporality, i.e. the order in which arguments are enunciated, so as to better model dialogues, reason about causal relationships between variables and provide explanations. Based on a formal method to rewrite the concepts of acyclic abstract argumentation frameworks in an action language, we discuss new ways to generate explanations based on a dynamic dialogue. As this modelling offers tools for visually narrating the evolution of the dialogue and the acceptability of its arguments, it is a good starting point for studying human-centred explanations. In addition, we investigate how richer causal relations can be used to provide explanations in addition to the graphical representation. Yet we observe that the extracted causal chains can be very long and contain redundancies, and thus do not necessarily provide satisfactory explanations. We therefore discuss directions to derive human-centred explanations from these causal chains satisfying recommendations provided by cognitive science studies.

Mots clés: Action Languages; Actual Causality; XAI; Argumentation

Domaines: Intelligence artificielle [cs.AI]

Yann Munro, Camilo Sarmiento, Isabelle Bloch, Gauvain Bourgne, Marie-Jeanne Lesot. Dynamic Argumentation and Action Languages: Towards Explanations. The Fourth Workshop on Explainable Logic-Based Knowledge Representation (XLoKR 2023), Sep 2023, Rhodes, Greece. ⟨hal-04179071⟩