Journal of E - Technology Jonathan Kobbe, Ioana Hulpus, Heiner Stuckenschmidt 15 2 2024 https://doi.org/10.6025/jet/2024/15/2/51-63 https://www.dline.info/jet/fulltext/v15n2/jetv15n2_1.pdf Classifying Arguments Argument relations classification is a way of classifying the type of relationship between two argument units. Current models mainly rely on surface-level language features such as discourse markers, modal, or adverbial to classify the relationship. However, a model that primarily relies on language features to classify an argument can be easily misled by the style rather than the content of the argument, particularly when a weak argument is masked by strong language. This paper examines the challenges and potential advantages of knowledge-based argument analysis in advancing the current state of argument analysis towards a deeper, knowledge-driven comprehension and representation of arguments. We propose an Arguments Classification System that uses linguistic and knowledge-based features to classify Arguments. We start with a Neural Baseline Model for classifying a Pair of Arguments based on the Siamese Network and expand it with a set of Features derived from two additional background knowledge sources: ConceptNet and DBpedia.