Artificial Intelligence (AI) holds significant promise for healthcare but often struggles to transition from development to clinical integration. This paper argues that Learning Health Systems (LHS)-socio-technical ecosystems designed for continuous data-driven improvement-provide a potential "glide path" for safe, sustainable AI deployment. Just as modern aviation depends on instrument landing systems, the safe and effective integration of AI into healthcare requires the socio-technical infrastructure of LHSs, that enable iterative development and monitoring of AI tools, integrating clinical, technical, and ethical considerations through stakeholder collaboration. They address key challenges in AI implementation, including model generalizability, workflow integration, and transparency, by embedding co-creation, real-world evaluation, and continuous learning into care processes. Unlike static deployments, LHSs support the dynamic evolution of AI systems, incorporating feedback and recalibration to mitigate performance drift and bias. Moreover, they embed governance and regulatory functions-clarifying accountability, supporting data and model provenance, and upholding FAIR (Findable, Accessible, Interoperable, Reusable) principles. LHSs also promote "human-in-the-loop" safety through structured studies of human-AI interaction and shared decision-making. The paper outlines practical steps to align AI with LHS frameworks, including investment in data infrastructure, continuous model monitoring, and fostering a learning culture. Embedding AI in LHSs transforms implementation from a one-time event into a sustained, evidence-based learning process that aligns innovation with clinical realities, ultimately advancing patient care, health equity, and system resilience. The arguments build on insights from an international workshop hosted in 2025, offering a strategic vision for the future of AI in healthcare.

Dialectical Classical Argumentation (Dialectical Cl-Arg) has been shown to satisfy rationality postulates under resource bounds.In particular, the consistency and non-contamination postulates are satisfied despite dropping the assumption of logical omniscience and the consistency and subset minimality checks on arguments' premises that are deployed by standard approaches to Cl-Arg.This paper studies Dialectical Cl-Arg's formalisation of Preferred Subtheories (PS) nonmonotonic reasoning under resource bounds.The contribution of this paper is twofold.First, we establish soundness and completeness for Dialectical Cl-Arg's credulous consequence relation under the preferred semantics and credulous PS consequences.This result paves the way for the use of argument game proof theories and dialogues that establish membership of arguments in admissible (and so preferred) extensions, and hence the credulous PS consequences of a belief base.Second, we refine the non-standard characteristic function for Dialectical Cl-Arg, and use this refined function to show soundness for Dialectical Cl-Arg consequences under the grounded semantics and resource-bounded sceptical PS consequence.We provide a counterexample that shows that completeness does not hold.However, we also show that the grounded consequences defined by Dialectical Cl-Arg strictly subsume the grounded consequences defined by standard Cl-Arg formalisations of PS, so that we recover sceptical PS consequences that one would intuitively expect to hold.

While there is universal agreement that agents ought to act ethically, there is no agreement as to what constitutes ethical behaviour. To address this problem, recent philosophical approaches to `moral uncertainty' propose aggregation of multiple ethical theories to guide agent behaviour. However, one of the foundational proposals for aggregation - Maximising Expected Choiceworthiness (MEC) - has been criticised as being vulnerable to fanaticism; the problem of an ethical theory dominating agent behaviour despite low credence (confidence) in said theory. Fanaticism thus undermines the `democratic' motivation for accommodating multiple ethical perspectives. The problem of fanaticism has not yet been mathematically defined. Representing moral uncertainty as an instance of social welfare aggregation, this paper contributes to the field of moral uncertainty by 1) formalising the problem of fanaticism as a property of social welfare functionals and 2) providing non-fanatical alternatives to MEC, i.e. Highest k-trimmed Mean and Highest Median.

This paper makes a proof-theoretic contribution to resource-bounded dialectical argumentation. Practical deployment of argumentation-based nonmonotonic reasoning can benefit from integration of proof-theoretic means for construction and evaluation of arguments, while accommodating agents with bounded resources. We present a nonmonotonic proof system that implements a generalization of dialectical argumentation, adopting arguments that differentiate between committed and supposed premises, while integrating rules for constructing arguments. The proof system adopts annotations to capture the changing status of arguments in a derivation and employs annotation revision rules that evaluate the dialectical acceptability of these arguments, yielding rational outcomes under resource bounds. Soundness and completeness is shown for the dialectical grounded semantics.

Dialectical Classical Logic Argumentation (D-Cl-Arg) formalises maxi-consistent non-monotonic reasoning under the practical assumption that agents have bounded resources for classical inference, and that agents do not typically check arguments’ premises for subset minimality and consistency. However, D-Cl-Arg still satisfies all rationality postulates. Moreover D-Cl-Arg accommodates uses of argument characteristic of dialectical practice. This paper extends D-Cl-Arg to accommodate further dialectical uses of argument; in particular unrestricted rebuts on the deductively derived conclusions of arguments, and Occam Razor defeats that dialectically demonstrate that an argument makes use of redundant premises. We show that all rationality postulates are still satisfied, while relaxing constraints on preference relations that were previously required to prove rationality.

Argument schemes represent stereotypical patterns of reasoning that capture the inferences from premise(s) to conclusion. Despite their usefulness in argument mining, argument scheme classification remains a largely understudied task in NLP. In this paper, we present EthiX, a novel dataset for classifying argument schemes, comprising arguments spanning 22 ethical topics which are manually annotated with argument schemes following Walton’s taxonomy. We evaluate pre-trained models fine-tuned on our dataset and propose a baseline to the community.

Conference paper presented at the Biennial International Conferences on Computational Models of Argument (COMMA), Perugia, Italy [Online], 8-11 September, 2020.

Dialogical generalisations of formal logic-based argumentation are typically restricted to a limited set of locutions e.g., assert, why, claim or prefer. However, the use of enthymemes (i.e., arguments with incomplete logical structure) warrant extending this set of locutions. This paper formalises the use of additional novel locutions that account for the use of enthymemes and are typical of real world dialogues. We thus close the gap between formal logic-based models of dialogue and the kinds of dialogue studied by the informal logic community, which focus on more human-oriented models of dialogue. This is important if formal models of dialogues are to provide normative support for human-human debate, as well as for enabling computational and human agents to jointly reason via dialogue.

ASPIC+ is an established general framework for argumentation and non-monotonic reasoning. However, ASPIC+ does not satisfy the non-contamination rationality postulates, and moreover, tacitly assumes unbounded resources when demonstrating satisfaction of the consistency postulates. In this paper we present a new version of ASPIC+ – Dialectial ASPIC+ – that is fully rational under resource bounds.

This paper discusses reasoning about norms using Extended Argumentation Frameworks. We argue that norms emerge from a process of value based practical reasoning. We model practical reasoning using Action Based Alternating Transition Systems with Values, with the resulting arguments evaluated in an Extended Argumentation Framework, which permits arguments to attack attacks as well as arguments, and so provides a natural way of representing exceptions. Following our consideration of how norms are justified, we consider their application. Our representation of norms in EAFs enables us to offer an accessible account of permissions and exceptions. We illustrate our approach with an example relating to some US Supreme Court Fourth Amendment cases.

The paper develops a formal theory of the degree of justification of arguments, which relies solely on the structure of an argumentation framework, and which can be successfully interfaced with approaches to instantiated argumentation. The theory is developed in three steps. First, the paper introduces a graded generalization of the two key notions underpinning Dung's semantics: self-defense and conflict-freeness. This leads to a natural generalization of Dung's semantics, whereby standard extensions are weakened or strengthened depending on the level of self-defense and conflict-freeness they meet. The paper investigates the fixpoint theory of these semantics, establishing existence results for them. Second, the paper shows how graded semantics readily provide an approach to argument rankings, offering a novel contribution to the recently growing research programme on ranking-based semantics. Third, this novel approach to argument ranking is applied and studied in the context of instantiated argumentation frameworks, and in so doing is shown to account for a simple form of accrual of arguments within the Dung paradigm. Finally, the theory is compared in detail with existing approaches.

In this demo paper, we propose an encoding for Metalevel Argumentation Frameworks (MAFs) to be used in Aspartix, an Answer Set Programming (ASP) approach to find the justified arguments of an AF. MAFs provide a uniform encoding of object level Dung Frameworks and extensions thereof that include values, preferences and attacks on attacks (EAFs). The justification status of arguments in the object level AF can then be evaluated and explained through evaluation of the arguments in the MAF. The demo includes multiple examples from the literature to show the applicability of our proposed encoding for translating various object level AFs to the uniform language of MAFs.

We directly instantiate metalevel argumentation frameworks (MAFs) to enable argumentation-based reasoning about information relevant to various applications. The advantage of this is that information that typically cannot be incorporated via the instantiation of object-level argumentation frameworks can now be incorporated, in particular information referencing (1) preferences over arguments, (2) the rationale for attacks, and (3) the dialectical effect of critical questions that shifts the burden of proof when posed. We achieve this by using a variant of ASPIC+ and a higher-order typed language that can reference object-level formulae and arguments. We illustrate these representational advantages with a running example from clinical decision support.

This short paper describes the design of the CONSULT system, a decision-support tool intended to help patients suffering from chronic conditions self-manage their health. The system takes input from multiple sources, including commercial wellness sensors and patient's electronic health record, to inform an intelligent back-end that reasons about day-to-day health management decisions, customised for individual patients. The architecture of the system features a modular structure for allowing input from a range of different sources, a reasoning engine underpinned by computational argumentation that constructs weighted opinions using these inputs and knowledge about their sources, and an interaction agent driven by argumentation-based dialogue that responds to user queries.

Classical logic argumentation (Cl-Arg) under the stable semantics yields argumentative characterisations of non-monotonic inference in Preferred Subtheories. This paper studies these characterisations under both the standard approach to Cl-Arg, and a recent dialectical approach that is provably rational under resource bounds. Two key contributions are made. Firstly, the preferred extensions are shown to coincide with the stable extensions. This means that algorithms and proof theories for the admissible semantics can now be used to decide credulous inference in Preferred Subtheories. Secondly, we show that as compared with the standard approach, the grounded semantics applied to the dialectical approach more closely approximates sceptical inference in Preferred Subtheories.

This paper proposes mechanisms for agents to model other agents' beliefs and arguments, thus enabling agents to anticipate their interlocutors' arguments in dialogues, which in turn facilitates strategising and the use of enthymemes. In contrast with existing works on “opponent modelling” that treat arguments as abstract entities, the likelihood that an interlocutor can construct an argument is derived from the likelihoods that it possesses the beliefs required to construct the argument. We therefore address how a modelling agent can quantify the certainty that its interlocutor possesses beliefs, based on the modeller's previous dialogues, and the membership of its interlocutor in communities.

Classical logic based argumentation (ClAr) characterises single agent non-monotonic reasoning and enables distributed non-monotonic reasoning amongst agents in dialogues. However, features of ClAr that have been shown sufficient to ensure satisfaction of rationality postulates, preclude their use by resource bounded agents reasoning individually, or dialectically in real-world dialogue. This paper provides a new formalisation of ClAr that is both suitable for such uses and satisfies the rationality postulates. We illustrate by providing a rational dialectical characterisation of Brewka's non-monotonic Preferred Subtheories defined under the assumption of restricted inferential capabilities.

There is an increasing need for norms to be embedded in technology as the widespread deployment of applications such as autonomous driving and warfare becomes ever closer. Current approaches to norms in multi-agent systems tend either to simply make prohibited actions unavailable, or to provide a set of rules (principles) which the agent is obliged to follow. We argue that both these approaches are inadequate: in order to meet unexpected situations agents must be capable of violating norms, when it is appropriate to do so. This in turn requires that agents be able to reason about what they should do without reference to the norms. One way to achieve this is to conduct value based reasoning using an argumentation scheme designed for practical reasoning. Such reasoning requires that agents have an acceptable set of values and an acceptable ordering on them. We discuss what might count as an acceptable ordering on values, and how such an ordering might be determined. Law breaking is illustrated through a simple road traffic example.

The work of Caminada & Amgoud presents two possible ways of satisfying the rationality postulates: one using restricted rebut, and one using unrestricted rebut. Subsequent work on ASPIC+ has extended the work of Caminada & Amgoud, for instance by allowing preferences over arguments. However, such extensions have utilised restricted rebut only. In the current paper, we make the case for unrestricted rebut, and provide a formalism (called ASPIC−) that implements preferences between the defeasible rules, in the context of unrestricted rebut while still satisfying the rationality postulates of Caminada & Amgoud.

Enthymemes, arguments with incomplete structure, are a ubiquitous feature of human communication and argumentation. This paper proposes a way of representing enthymemes and arguments within the ASPIC+ framework, and how enthymemes are constructed based on estimates of shared knowledge. It then proposes a framework in which agents are capable of constructing a model of other agent's knowledge.

The Argument Interchange Format (AIF) has been devised in order to support the interchange of ideas and data between different projects and applications in the area of computational argumentation. In order to support such interchange, an abstract ontology for argumentation is presented, which serves as an interlingua between various more concrete argumentation languages. In this article, we aim to give what is essentially a logical specification of the AIF ontology by mapping the ontology onto the logical ASPIC+ framework for argumentation. We thus lay foundations for interrelating formal logic-based approaches to argumentation captured by the ASPIC+ framework and the wider class of argumentation languages, including those that are more informal and user-orientated.

Recently resolution of attacks has been studied in the context of abstract argumentation frameworks. In this paper it is claimed that resolutions should be studied under the assumption that they are generated through the acquisition of preference information, and that this implies that the existing study of resolutions has limited applicability. A formalisation of preference-based resolutions is defined in the context of the ASPIC+framework for structured argumentation, and several properties of resolutions are proven or disproven. It is also argued that when resolutions are modelled without specifying the structure of arguments, then it is easy to overlook that assumptions made at the abstract level do not hold for all reasonable instantiations of the abstract framework, thus reducing its significance.

The ASPIC+framework is a general framework for argumentation-based inference which aims to unifies two research strands: those in which arguments can only be attacked on their defeasible inferences and those in which arguments can only be attacked on their premises. The framework is meant to define a wide class of instantiations of abstract argumentation frameworks and to support the investigation of rationality postulates for argumentation-based inference. Recently, it has been argued that the ASPIC+framework suffers from several weaknesses. In this paper these criticisms are argued to be based on a number of misconceptions on the nature of the ASPIC+framework.

S. Modgil, F. Toni; Special Issue on Argumentation in Agreement Technologies, Journal of Logic and Computation, Volume 22, Issue 5, 1 October 2012, Pages 9

The abstract nature of Dung’s theory of argumentation accounts for its wide-spread application as a general framework for various species of non-monotonic reasoning, and, more generally, reasoning in the presence of conflict. In this pa-per we formalise reasoning about argumentation within the Dung argumentation paradigm itself. A metalevel Dung argumentation framework is itself instantiated by arguments that make statements about arguments, their interactions, and their evaluation in an object-level argumentation framework. We show how Dung’s theory, and object level extensions of Dung’s theory, such as those intended to accommodate preferences, can then be uniformly characterised by metalevel argu-mentation in a Dung framework. We then discuss how this provides for application of the full range of theoretical and practical developments of Dung’s theory, to ex-tensions of Dung’s theory, and provides for integration and further augmentation of these extensions.

This paper argues that accrual should be modelled in terms of reasoning about the application of preferences to sets of arguments, and shows how such reasoning can be formalised within metalevel argumentation frameworks. These frameworks adopt the same machinery and level of abstraction as Dung's argumentation framework. We thus provide a dialectical argumentation semantics that integrates accrual, and illustrate our approach by instantiating our framework with the arguments and attacks defined by an object level formalism that accommodates reasoning about priorities over sets of rules.

This paper combines two recent extensions of Dung's abstract argumentation frameworks in order to define an abstract formalism for reasoning about preferences in structured argumentation frameworks. First, extended argumentation frameworks extend Dung frameworks with attacks on attacks, thus providing an abstract dialectical semantics that accommodates argumentation-based reasoning about preferences over arguments. Second, a recent extension of the ASPIC framework (ASPIC+) instantiates Dung frameworks with accounts of the structure of arguments, the nature of attack and the use of preferences to resolve attacks. In this paper, ASPIC+ is further developed in order to define attacks on attacks, resulting in a dialectical semantics that accommodates argumentation based reasoning about preferences in structured argumentation. Then, some recently proposed rationality postulates for structured extended argumentation are proven to hold.

In this paper we discuss how recent developments in argumenta-tion frameworks, most notably Extended Argumentation Frame-works, can inform the representation of a body of case law using abstract argumentation techniques. This builds on previous work which has first used abstract Argumentation Frameworks, and then Value based Argumentation Frameworks for this purpose. Extended Argumentation Frameworks augment Argumentation Frameworks to not only allow arguments to be attacked, but also attacks to be attacked. This allows argumentation based reason-ing about information normally assumed to be metalevel to the ob-ject level domain of argumentation, including argumentation over preferences, values and the audience based ranking of values pro-moted by arguments. The Extended Argumentation Frameworks can then be rewritten as standard Argumentation Frameworks, so that cases, and values and their rankings relevant to the cases, can be reasoned about using standard dialogue games for Argumenta-tion Frameworks. In this way precedents can be represented as col-lections of arguments and dialogues using these arguments. Now, when confronted with a new case, these dialogues may be used to identify ways of deploying the arguments in the new case so as to reach a favourable position. 1.