Abstract
This paper presents a framework to convert argumentative texts into argumentknowledge graphs (AKG). Starting with basic annotations of argumentativecomponents (ACs) and argumentative relations (ARs), we enrich the informationby constructing a knowledge base (KB) graph with metadata attributes for nodes.Next, we use premises and inference rules from the KB to form arguments byapplying modus ponens. From these arguments, we create an AKG. The nodes andedges of the AKG have attributes that capture important argumentative features.We also find missing inference rules by identifying markers. This makes itpossible to identify undercut attacks that were previously undetectable inexisting datasets. The AKG gives a graphical view of the argumentativestructure that is easier to understand than theoretical formats. It alsoprepares the ground for future reasoning tasks, including checking thecoherence of arguments and identifying opportunities for revision. For this, itis important to find indirect relations, many of which are implicit. Ourproposed AKG format, with annotated inference rules and modus ponens, will helpreasoning models learn the implicit indirect relations that require inferenceover arguments and the relations between them.