A bipolar argumentation-based decision framework
Résumé
The paper emphasizes the bipolar nature of the evaluation of decision results by making an explicit dis-tinction between prioritized goals to be pursued, and prioritized rejec-tions that are stumbling blocks to be avoided. This is the basis for an ar-gumentative framework for decision. Each decision is supported by argu-ments emphasizing its positive con-sequences in terms of goals satisfied, or rejections avoided by that deci-sion. A decision can also be attacked by arguments emphasizing its nega-tive consequences in terms of missed goals, or rejections reached by that decision. Thus, such bipolar setting provides a richer typology of arguments, and consequently rich principles for com-paring decisions. We will show also that such a setting allows not only comparing decisions, but also defining the status of each deci-sion. Four main status are dis-tinguished: recommended decisions, discommended decisions, controver-sial decisions and finally neutral de-cisions. The paper presents a unified logical argumentation-based model in which different decision processes (decision under uncertainty, multiple criteria decision, and rule-based decision) are captured.
Domaines
Intelligence artificielle [cs.AI]Origine | Fichiers produits par l'(les) auteur(s) |
---|