Learning Argumentation Frameworks from Labelings[zur Übersicht]
Since its introduction by Dung in 1995, a central aspect of abstract argumentation has been the argumentation semantics. These semantics are concerned with computing sets of arguments which satisfy certain criteria based on the structure of attacks in the argumentation framework.Two approaches can be distinguished for the output of argumentation semantics: extensions and labelings. In the scope of this work we will mainly focus on labeling-based argumentation semantics.A topic which has not received as much attention so far is going in the opposite direction of semantics: Constructing argumentation frameworks from a given set of labelings such that these argumentation frameworks produce all input labelings. This master thesis presents a novel approach for learning argumentation frameworks from labelings. An important element of this approach is computing all argumentation frameworks that satisfy the input labelings instead of simply finding any suitable argumentation framework. This is especially important, for example, if we receive additional labelings at a later time and want to refine our result without having to start all over again. The developed algorithm will be compared to the existing work on this topic and an evaluation of its performance has been conducted.
30.09.21 - 10:15
via Big Blue Button