Modelling online debates with argumentation theory

Abstract

It is important to study and understand Internet debates because they often have consequences in the offline world, for better or worse. We show that argumentation theory, a branch of AI concerned with the resolution of disagreements, provides a powerful toolbox with which we can represent and reason about such debates. After summarising the relevant ideas of argumentation theory, we overview three recent contributions from the authors and their collaborators: (1) on how to automatically identify reply polarity (agreement or disagreement) between arguments submitted in online debates, (2) on locating where the justified arguments are likely to be and how that depends on the "degree of antagonism" of the debate, and (3) on how to present the arguments made in debates such that a reader would get as many of the justified arguments as possible without having to read the entire debate. We hope that this will lead to further work that applies argumentation theory to model and analyse online debates.