Jinfeng Zhong: Argumentation meets matrix factorization: A dual perspective for explainable recommendations

Factorization-based models gained prominence during the Netflix Challenge (2007) and have since demonstrated strong performance in predicting user ratings. However, their limited interpretability often hinders users from understanding the rationale behind recommendations. In contrast, argumentation-based methods offer a different perspective: they model human-like reasoning by structuring information as arguments and counterarguments. They excel in explainability but typically fall short in accuracy. To address this trade-off, we propose a novel framework, Context-Aware Feature-Attribution Through Argumentation (CA-FATA), which combines the predictive power of matrix factorization with the interpretability of argumentation frameworks. In CA-FATA, each user-item interaction is modeled using an argumentation framework. Item features are represented as arguments, and users' ratings determine the arguments' strengths. Additionally, the model incorporates users' contextual information (e.g., time, location) to further improve predictive performance. Empirical evaluations on real-world datasets show that CA-FATA excels in predictive accuracy and interpretability. It outperforms existing argumentation-based methods and achieves comparable results with state-of-the-art context-free and context-aware models. CA-FATA also supports multiple explanation formats, including template-based explanations, interactive feedback, and contrastive reasoning. Furthermore, it alleviates the cold-start problem by clustering users based on feature preferences.

Katie Atkinson: Reflections on modelling reasoning in legal cases

Computational models of argument have been demonstrated to be able to model reasoning about legal cases, using a variety of different approaches. Nonetheless, scaling up the symbolic models can be challenging and in light of this, data-driven techniques have recently received a lot of attention as a way to conduct the task of ‘predicting’ the outcomes of legal cases, however these methods often diminish the focus on capturing reasoning. In this talk I will provide some reflections from a variety of exercises undertaken to use computational models of argument and machine learning techniques to tackle the variety of tasks involved in producing decision-support tools for supporting reasoning about legal cases, reporting on recent work aimed at combining the two approaches. The techniques and experiments undertaken will be demonstrated through cases studies undertaken on real world domains, most notably the European Convention on Human Rights.

Nir Oren and Bruno Yun: Exploring Inverse Problems for Gradual Semantics in Argumentation

In this talk we consider several aspects of inverse problems under gradual semantics. While traditional argumentation reasoning proceeds from inputs (e.g., initial argument weights and attack relations) to outputs (e.g., acceptability degrees), we reverse this process to explore how some given outputs can be used to reconstruct the inputs. We consider two main problems, namely inferring initial argument weights that lead to a specified set of acceptability degrees and identifying attack relations that realise a target ranking. Finally, we introduce an application domain which utilises some of these techniques, demonstrating how users can be supported in refining their beliefs using intervals and ensuring that these beliefs are rational.

Liuwen Yu: Conceptualizations and challenges for distributed argumentation technology     

In this talk, I will discuss the foundations and challenges of Distributed Argumentation Technology (DAT), a computational approach that incorporates argumentation reasoning mechanisms within multi-agent systems. I will begin by illustrating the relevance of DAT through its applications, including Intelligent Human-Input-Based Blockchain Oracles, agents based on large language models (LLMs), discretionary decision-making, and regulatory compliance in finance. Each application calls for distinct models of formal argumentation, leading to the central focus of this talk: the foundations of formal argumentation. I distinguish three conceptualizations—argumentation as balancing, argumentation as dialogue, and argumentation as inference— representing distinct viewpoints on argumentation from stakeholders. Finally, I will discuss open challenges for DAT, drawing from our recent chapter in the third volume of the Handbook of Formal Argumentation:

https://yuliuwen.github.io/assets/files/HOFA3_Chapter_Thirteen_Challenges.pdf.

Paola Budan: Formalization of Argumentation Schemes with Abstract and Structured Argumentation

We will discuss the state of the art and some research possibilities on the research line that focuses on making argumentation schemes computable devices. First, we will share preliminary work on extending argumentation schemes (AS) in light of critical thinking and their potential implementation in DeLP. Then, we will focus on AS based on analogies as tools for detecting and characterizing communities. Finally, we will address AS based on Popular Opinion to characterize, together with Labelled Argumentation Frameworks, different arguments within a community.

Timotheus Kampik: Change in quantitative bipolar argumentation: Sufficient, necessary, and counterfactual explanations

In non-monotonic reasoning and (more specifically) in formal argumentation, the ability to change one's conclusions is a crucial feature. With the rise of interest in explainable reasoning, the question of why an inference has changed -- why we infer B and no longer A -- can be considered of central relevance. To shed light on this question, I present our recent paper introducing a formal approach to explaining change of inference in Quantitative Bipolar Argumentation Frameworks (QBAFs). When drawing conclusions from a QBAF and updating the QBAF to then again draw conclusions (and so on), the approach traces changes – called strength inconsistencies – in the partial order over argument strengths that a semantics establishes on two arguments of interest, called topic arguments. We trace the causes of strength inconsistencies to specific arguments, which then serve as explanations. We identify sufficient, necessary, and counterfactual explanations for strength inconsistencies and show that strength inconsistency explanations exist if and only if an update leads to strength inconsistency. Finally, we define a heuristic-based approach to facilitate the search for strength inconsistency explanations. The approach is implemented as a C library with Python bindings (https://github.com/TimKam/Quantitative-Bipolar-Argumentation), facilitating usage in prototyping and data science contexts.

Bart Verheij: Artificial Intelligence and Argumentation

For long there has been a fruitful interaction between research in Artificial Intelligence and in Argumentation Theory. For instance, John Pollock simultaneously worked on the theory of argumentation and on the implementation of an artificial rational agent.. He spoke of `how to build a person' and the `grand problem of human level intelligence', and even gave his artificial person a name: OSCAR. Phan Minh Dung's paper on argumentation frameworks starts with the idea that `argumentation constitutes a major component of human intelligence', and has the purpose `to study the fundamental mechanism, humans use in argumentation, and to explore ways to implement this mechanism on computers'. The paper was published in the journal Artificial Intelligence and won the Artificial Intelligence Journal Classic Paper Award (announced at IJCAI 2018). 

Also in these days of data-driven artificial intelligence, an argumentation perspective has much to offer. For instance, in a 2017 lecture `Arguments for Good Artificial Intelligence', I explained that argumentation is needed for the design of a responsible AI by bridging knowledge-based and data-driven approaches to AI. In the talk, I present an update on recent research performed on AI and argumentation in Groningen, with a focus on connections between knowledge, data and reasoning using an argumentation perspective.

Guillermo Simari: Focusing the Argumentative Procces: Neighborhood-Based Semantics in Abstract Argumentation

The introduction of abstract argumentation has allowed the study of many exciting characteristics of the argumentation process. Nevertheless, while helpful in many aspects, abstraction diminishes the knowledge representation capabilities available to describe naturally occurring features of argumentative dialogues. One of these elements is the consideration of the topics involved in a discussion. In studying dialogical processes, participants recognize that some topics are closely related to the original issue. In contrast, others are more distant from the central subject or refer to entirely unrelated matters. Consequently, it is reasonable to study several different argumentation semantics that consider a discussion's focus to evaluate acceptability. In this talk, we will introduce the necessary representational elements required to reflect the focus of a debate, and we will propose an extension of the semantics for multi-topic abstract argumentation frameworks acknowledging that every argument has its own "zone of relevance" in the argumentation framework, leading to a concept of a neighborhood of legitimate defenses.

Matthias Thimm: Revisiting initial sets in abstract argumentation

I revisit the notion of initial sets by Xu and Cayrol, i.e., non-empty minimal admissible sets in abstract argumentation frameworks. Initial sets are a simple concept for analysing conflicts in an abstract argumentation framework and to explain why certain arguments can be accepted. I contribute with new insights on the structure of initial sets and devise a simple non-deterministic construction principle for any admissible set, based on iterative selection of initial sets of the original framework and its induced reducts. In particular, I characterise many existing admissibility-based semantics via this construction principle, thus providing a constructive explanation on the structure of extensions. I conclude with some comments on computational complexity.

Wolfgang Dvořák, Anna Rapberger, Stefan Woltran: A claim-centric view on abstract argumentation

Abstract argumentation frameworks have been introduced by Dung as part of an argumentation process, where arguments and conflicts are derived from a given knowledge base. It is solely this relation between arguments that is then used in order to identify acceptable sets of arguments. A final step concerns the acceptance status of particular statements by reviewing the actual contents of the acceptable arguments. However, a certain subtlety - which is often neglected -- occurs when different arguments with the same claim are constructed during this process. 

On the one hand, it makes the actual notion of acceptance problems ambiguous - a claim-centric definition of acceptance leads to changes compared to the known complexity landscape of abstract argumentation. On the other hand, the additional layer of claims offers several possibilities of lifting concepts of argumentation semantics to claim-based semantics - we present two natural variants and compare them in terms of computational complexity, expressibility and I-maximality. 

We furthermore study under which circumstances the construction of multiple arguments having the same claim can be avoided, i.e. we investigate whether the abstraction step can be done in such a way that each argument represents a unique claim. We show that scenarios where arguments with the same claim have the same outgoing attacks can be equivalently represented as argumentation frameworks with collective attacks but unique claims.

Jérémie Dauphin: A Structured Argumentation Framework for Modeling Debates in the Formal Sciences

Scientific research in the formal sciences comes in multiple degrees of formality: fully formal work; rigorous proofs that practitioners know to be formalizable in principle; and informal work like rough proof sketches and considerations about the advantages and disadvantages of various formal systems. This informal work includes informal and semi-formal debates between formal scientists, e.g. about the acceptability of foundational principles and proposed axiomatizations. In this talk, we propose to use the methodology of structured argumentation theory to produce a formal model of such informal and semi-formal debates in the formal sciences. For this purpose, we propose ASPIC-END, an adaptation of the structured argumentation framework ASPIC+ which can incorporate natural deduction style arguments and explanations. We illustrate the applicability of the framework to debates in the formal sciences by presenting a simple model of some arguments about proposed solutions to the Liar paradox, and by discussing a more extensive - but still preliminary - model of parts of the debate that mathematicians had about the Axiom of Choice in the early 20th century.

Davide Grossi: Deliberative Consensus

In this talk, I will step a bit outside the standard formal argumentation models to address the topic of democratic deliberation. The talk focuses on a setting in which a community wishes to identify a strongly supported proposal from a large space of alternatives, in order to change the status quo. I will describe a deliberation process in which agents dynamically form coalitions around proposals that they prefer over the status quo. Using this model I will show how the properties of the underlying abstract space of proposals and the ways in which agents can form coalitions affect the success of deliberation in creating consensus. We show that, as the complexity of the proposal space increases, more complex forms of coalition formation are required in order to guarantee success. Intuitively, this seems to suggest that complex deliberative spaces require more sophisticated coalition formation abilities on the side of the agents. The model provides theoretical foundations for the analysis of deliberative processes in systems for democratic deliberation support, such as Polis or LiquidFeedback. This is joint work with Edit Elkind (University of Oxford), Ehud Shapiro (Weizmann Institute) and Nimrod Talmon (Ben-Gurion University)

Ringo Baumann: Modularization and Weak Admissibility

In this talk I will highlight the most central results regarding weak admissibility and modularization. The former was recently introduced to overcome an acceptability problem arising in presence of self-defeating arguments underlying all admissibility-based semantics. The major difference between the standard definition of admissibility and our new one is that arguments do not have to defend themselves against all attackers. Modularization is an abstract property of semantics describing how to obtain further extensions given an initial one. Its importance for the study of abstract argumentation theory is shown by its ability to alternatively characterize classical and non-classical semantics.

Anthony Hunter: Towards Computational Persuasion for Behaviour Change Applications

The aim of behaviour change is to help people to change aspects of their behaviour for the better (e.g., to decrease calorie intake, to drink in moderation, to take more exercise, to complete a course of antibiotics once started, etc.). Recent developments in computational modelling of argument (a subfield of AI) are leading to technology for persuasion that can potentially be harnessed in behaviour change applications. Using this technology, a software system and a user can exchange arguments in a dialogue. So the system gains information about the user’s perspective, provides arguments to fill gaps in the user’s knowledge, and attempts to overturn misconceptions held by the user.  Our work has focused on modelling the beliefs and concerns of the user, and harnessing these to make the best choices of move during the dialogue for persuading the user to change their behaviour.  During this talk, the background to, and components of, our approach will be presented together with some promising preliminary results with participants.

Serena Villata: Artificial Argumentation for Humans

In this talk, I will present existing approaches in the area of Argument Mining. In order to cut in on a debate on the web, the participants need first to evaluate the opinions of the other users to detect whether they are in favor or against the debated issue. Bipolar argumentation proposes algorithms and semantics to evaluate the set of accepted arguments, given the support and the attack relations among them. Therefore, an automated framework to detect the relations among the arguments represented by the natural language formulation of the users’ opinions is needed. My talk addresses this open issue by proposing and evaluating the use of natural language processing methods to identify the arguments and predict the relations between them. Three scenarios will be discussed, namely social media, clinical trials and political debates. I will conclude with an overview about the open challenges in the Argument Mining research area. 

Paola Daniela Budan: Similarity notions in Bipolar Argumentation

In this talk we address a  model to calculate the similarity degree between arguments, and to evaluate the support and attack relations between them. In this direction, we present a novel mechanism to determine the similarity between two arguments based on descriptors that represent particular aspects associated with these arguments. This mechanism involves a comparison process influenced by the context in which the process develops, where this context provides the relevant aspects that need to be analyzed in the application domain. Then, we use this similarity measure as a quantity to compute the result of attacks and supports in the argumentation process.  Later, we present a Similarity Bipolar Argumentation Framework, in which the sets of admissible arguments are determined based on the similarity applied to the relations between arguments as a result of carrying out a semantic evaluation associated with the arguments in the argumentative process.

Henry Prakken: A Top-level Model of Case-based Argumentation for Explanation

In this talk I will outline a formal top-level model of explaining the outputs of machine-learning-based decision-making applications. The model draws on  AI & law research on argumentation with cases, which models how lawyers draw analogies to past cases and discuss their relevant similarities and differences in terms of relevant factors and dimensions in the problem domain.  A case-based approach is natural since the input data of machine-learning applications can be seen as cases. While the approach is motivated by legal decision making, it also applies to other kinds of decision making, such as commercial decisions about loan applications or employee hiring, as long as the outcome is binary and the input conforms to the model's factor- or dimension format. The model is top-level in that it can be extended with more refined accounts of similarities and differences between cases. I will first outline the formal model (which is based on the work published at DEXA HAI 2020) and then discuss some experiments done by Rosa Ratsma in which the model was applied to several machine-learning-based decision-making applications.

Sanjay Modgil: Dialectical Formalizations of Non-monotonic Reasoning: Rationality under Resource Bounds 

Argumentative formalizations of non-monotonic logics enable distributed reasoning amongst multiple agents. Recent developments in AI suggest that these dialogical/dialectical models of non-monotonic reasoning may play an important role in solving the value loading/alignment problem. Moreover, recent work in cognitive science suggests that the normative use of logic in guiding human reasoning  is better realised through such dialogical models. ASPIC+ — a widely used framework for defining dialectical formalizations of non-monotonic logics — provides guidelines for ensuring that such formalizations yield rational outcomes. However, rationality is guaranteed only under the assumption that agents have unbounded resources. Moreover, ASPIC+ is not fully rational, in the sense that given a set of formulae A, then the argumentation defined inferences from A may be retracted when adding some set B that is syntactically disjoint from A (a problem known as contamination). In this talk I will present a new version of ASPIC+ that adopts a radically new notion of argument and counter-argument, and that can be shown to be fully rational under resource bounds. This then means that resource bounded agents can now reason non-monotonically, as individuals or jointly via dialogue, while guaranteeing rational outcomes.

Francesca Toni: Argumentative explanations for recommendations

Explainable AI (XAI) has recently emerged as a popular topic in AI. In this talk I will argue that computational argumentation is uniquely well-placed to support XAI, especially as a conduit for information exchange between AI systems and humans.  I will corroborate my arguments by showing how argumentation can support explainable recommender systems in two ways. In both illustrations, explanations are drawn from automatically extracted argumentation frameworks. In the first recommender system, these argumentation frameworks are mined from a conventional hybrid recommender system, whereas in the second they are drawn from data. Finally, I will show how extracted argumentation frameworks can support a variety of explanation contents and styles, suitable to a variety of users.

Leon van der Torre: The Principle-Based Approach to Abstract Argumentation Semantics

The principle-based or axiomatic approach to abstract argumentation semantics has been used to choose an argumentation semantics for a particular application, to guide the search for new argumentation semantics, to relate abstract argumentation to other formal theories, and in the design of algorithms. In this talk I discuss motivations, results, and open problems.

Anthony Hunter: Epistemic Graphs for Representing and Reasoning with Positive and Negative Influences of Arguments

This paper introduces epistemic graphs as a generalization of the epistemic approach to probabilistic argumentation. In these graphs, an argument can be believed or disbelieved up to a given degree, thus providing a more fine--grained alternative to the standard Dung's approaches when it comes to determining the status of a given argument. Furthermore, the flexibility of the epistemic approach allows us to both model the rationale behind the existing semantics as well as completely deviate from them when required. Epistemic graphs can model both attack and support as well as relations that are neither support nor attack. The way other arguments influence a given argument is expressed by the epistemic constraints that can restrict the belief we have in an argument with a varying degree of specificity. The fact that we can specify the rules under which arguments should be evaluated and we can include constraints between unrelated arguments permits the framework to be more context--sensitive. It also allows for better modelling of imperfect agents, which can be important in multi--agent applications.

Pietro Baroni: Towards layered and modular abstractions in argumentation

The talk will discuss, with an exploratory perspective, some general issues about the use of abstraction in formal argumentation. Drawing analogies with other areas of computer science and referring to some recent contributions in the literature, it will be argued that there is room (and probably need) for further developments aimed at providing increasingly accurate and practically useful representations of the argumentation phenomenon. In particular, some preliminary ideas about investigating a modelling framework featuring different layers of abstractions will be discussed and some hints about exploiting modularity in this context will be provided.