Argumentation and Machine Learning

Aug 2, 2026

This chapter surveys research works that cross-fertilize Computational Argumentation and Machine Learning. A literature review identifies two main interaction purposes: argumentation for machine learning and machine learning for argumentation. The survey systematically compares works across dimensions such as the learning type and the argumentation framework form. It further categorizes interaction styles as synergistic, segmented, and approximated approaches, and analyzes which argumentation forms best support particular machine learning types and vice versa, while also discussing limitations and challenges for continued progress in AI.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages44

Reading Time1 h 51 min

AccessFree to download

Introduction

Outlook

Interaction Types (Synergistic, Segmented, Approximated)

arXiv :2410 .23724v 1 [ cs .AI] 31 Oct 2024 ARGUMENTATION AND MACHINE LEARNING ANTONIO RAGO* Department of Computing, Imperial College London [a.rago@imperial.ac.uk](a.rago@imperial.ac.uk) 􀀐 KRISTIJONAS CYRAS􀀃 Ericsson [kristijonas.cyras@ericsson.com](kristijonas.cyras@ericsson.com) JACK MUMFORD Department of Computer Science, University of Liverpool [jack.mumford@liverpool.ac.uk](jack.mumford@liverpool.ac.uk) OANA COCARASCU Department of Informatics, King’s College London [oana.cocarascu@kcl.ac.uk](oana.cocarascu@kcl.ac.uk) Abstract This chapter provides an overview of research works that present approaches with some degree of cross-fertilisation between Computational Argumentation and Machine Learning. Our review of the literature identi􀀂ed two broad themes representing the purpose of the interaction between these two areas: argumentation for machine learning and machine learning for argumentation. Across these two themes, we systematically evaluate the spectrum of works across various dimensions, including the type of learning and the form of argumentation framework used. Further, we identify three types of interaction between these two areas: synergistic approaches, where the Argumentation and Machine Learning components are tightly integrated; segmented approaches, where the two are interleaved such that the outputs of one are the inputs of the other; and approximated approaches, where one component shadows the other at *Equal contribution. 􀀐 RAGO , CYRAS, MUMFORD & COCARASCU a chosen level of detail. We draw conclusions about the suitability of certain forms of Argumentation for supporting certain types of Machine Learning, and vice versa, with clear patterns emerging from the review. Whilst the reviewed works provide inspiration for successfully combining the two 􀀂elds of research, we also identify and discuss limitations and challenges that ought to be addressed in order to ensure that they remain a fruitful pairing as AI advances. 1 Introduction In this chapter, we overview research works that combine Computational Argumentation (henceforth simply argumentation) and Machine Learning (ML) . In this section, we start with a general outlook of the themes and trends prevalent among the individual works overviewed in the chapter, including a loose categorisation of three types of interactions between ML an argumentation models, before brie􀀃y covering related work and papers we chose to omit. We then move on to our literature overview in §2 which we split by broad themes of the purposes of the interactions between ML and argumentation, brie􀀃y discussing promising research avenues. Finally, we conclude in §3 . We assume the reader to be familiar with forms and frameworks of argumentation, e.g. as discussed in the 1st Volume of the Handbook of Formal Argumentation (Baroni et al., 2018), and fundamentals of and common methods in ML, as easily accessible e.g. on Wikipedia. 1.1 Outlook Studies that consider interactions of argumentation and ML exhibit the following trends. Starting with ML characteristics, the type of learning employed is largely supervised, though there are instances of works that consider unsupervised as well as both un/supervised types of learning, and a reasonable amount of works that focus on Reinforcement Learning (RL) . In terms of ML models and algorithms employed, it is typical, but not exclusively so, to make use of simpler techniques and architectures, such as rule learning (including both rule induction and rule extraction), tree-based models (such as decision trees (DTs) and random forests (RFs)), naive Bayesian classi􀀂ers (NBCs), support vector machines (SVMs), (usually shallow and/or feed-forward) neural networks (NNs) and model-based RL. Some recent works though opt for modern com- ARGUMENTATION AND MACHINE LEARNING plex models, in particular graph neural networks (GNNs), typically available off-the-shelf. The data that is used in experiments is

FAQs

What are the two main purposes of connecting computational argumentation and machine learning described in the chapter?

The chapter distinguishes argumentation for machine learning and machine learning for argumentation, framing two broad themes for the interaction between the fields.

How does the chapter categorize the interaction styles between argumentation and machine learning?

It identifies three types: synergistic approaches with tight integration, segmented approaches where the components are interleaved, and approximated approaches where one component shadows the other at a chosen level of detail.

What common trends does the chapter report about machine learning methods used in the reviewed works?

Most works use supervised learning, with additional cases including unsupervised/both un-/supervised settings and some focus on reinforcement learning, often employing rule learning, tree-based models, Naive Bayes, SVMs, and sometimes neural networks or graph neural networks.

Aug 2, 2026

Uploaded by Seraphina