Graded Justification of Arguments via Internal and External Endogenous Features

摘要

We propose a framework to compute a graded justification of arguments and a ranking of them. The framework is based on two different features that can be directly extracted from an Argumentation Framework (endogenously). Hence, the suggested approach does not consider any side-information on arguments or attacks, e.g., in the form of preferences. The two features are derived from (i) allowing a number of attacks inside an extension, and (ii) computing how well such an extension can defend its arguments (the difference between the number of attacks and counter-attacks). The ranking of arguments is provided by computing their justification status w.r.t the semantics redefined through i and ii.