Assumption-Based Argumentation for Decision-Making with Preferences: A Medical Case Study

We present a formal decision-making framework, where decisions have multiple attributes and meet goals, and preferences are defined over individual goals and sets of goals. We define decision functions to select 'good' decisions according to an underlying decision criteria. We also define an argumentation-based computational mechanism to compute and explain 'good' decisions. We draw connections between decision-making and argumentation semantics: 'good' decisions are admissible arguments in a corresponding argumentation framework. To show the applicability of our approach, we use medical literature selection as a case study. For a given patient description, we select the most relevant medical papers from the medical literature and explain the selection.