Constraint-based causal discovery is brittle in finite-sample regimes because erroneous conditional-independence (CI) decisions can cascade into substantial structural errors. We propose Quantitative Argumentation for Causal Discovery (QACD), a semantics-driven framework that represents CI outcomes as graded, defeasible arguments rather than irreversible constraints. QACD maps statistical test outcomes to argument strengths and aggregates conflicting evidence through connectivity-mediated witness propagation, producing a fixed-point acceptability labeling over candidate adjacencies. Experiments on standard benchmark Bayesian networks suggest that QACD improves structural coherence and interventional reliability in several noisy or inconsistent CI regimes, while remaining competitive with classical constraint-based, hybrid, and prior argumentation-based baselines.

Large language models (LLMs) excel at complex reasoning and achieve human-like performance in many natural language processing tasks. However, they still struggle to reason effectively when faced with inconsistent or contradictory information. This capability gap raises significant concerns for real-world applications where reliable decision-making depends on reconciling conflicting evidence, such as legal analysis, medical diagnosis, and commonsense reasoning. In this paper, we focus on defeasible reasoning in natural language, a task that challenges LLMs to handle and resolve contradictory information. To improve the defeasible reasoning capability of LLMs, we propose LLM-ASPIC+, a framework combining neural language understanding with formal argumentation. Our framework harnesses LLMs’ capacity for grounding and contextual reasoning while integrating formal argumentation frameworks to establish systematic conflict resolution mechanisms lacking in LLMs. We also create MineQA, a newly synthesized dataset designed to evaluate multi-step defeasible reasoning under both strict and defeasible rules. LLM-ASPIC+ achieves state-of-the-art results on multi-step defeasible reasoning, with 87.1% accuracy on BoardGameQA-2 and 82.6% on BoardGameQA-3. These results show that integrating neural language models with formal argumentation effectively supports defeasible reasoning in natural language.

Mathematicians have long employed decomposition techniques to prove inequalities, yet automating this process remains a significant challenge in computational mathematics. We introduce IneqSearch, a hybrid reasoning system that integrates symbolic computation with large language models (LLMs) to address this challenge. IneqSearch reformulates inequality proving as a structured search problem: identifying appropriate combinations of theorems that decompose expressions into non-negative components. The system combines a symbolic solver for deductive reasoning with an LLM-based agent for constructive proof exploration, effectively implementing methodologies observed in formal mathematical practice. A key contribution of IneqSearch is its iterative learning mechanism that systematically incorporates newly proven results into its theorem database, enabling knowledge acquisition during practice that enhances its capabilities without requiring human intervention. In empirical evaluation on 437 Olympiad-level inequalities, IneqSearch successfully proves 342 problems, significantly outperforming existing methods and demonstrating the effectiveness of integrating symbolic and neural approaches for mathematical reasoning.

Graph Neural Networks (GNNs) have achieved remarkable success in node classification tasks on individual graphs. However, existing GNNs trained within a specific domain ( a.k.a., source domain) frequently exhibit unsatisfied performance when transferred to another domain ( a.k.a., target domain), due to the domain gap. To tackle this issue, Few Shot Graph Domain Adaptation (FSGDA) is introduced to the node classification task, facilitating knowledge transfer from a fully labeled source graph to a target graph with minimal annotations for each class. An intuitive solution is directly training the GNN with labeled source and target samples together. Nevertheless, there are two issues in this procedure: (1) When the annotations on the target domain used for training are extremely sparse, the GNN performance may significantly be damaged by nodes with the source-domain bias not aligning with the target-domain distribution. (2) Apart from the biased nodes, the low-value nodes among the remaining nodes impede the GNN learning for the core nodes, like the limited target training nodes. To address the above issues, we propose a new method for FSGDA, named GraphInflu, whose core idea is to grasp the key takeaways from the source domain to facilitate the adaptation process. It contains two characteristic modules, including the Supportive Node Selector and the Soft Logic-Inspired Node Reweighting. The former aims to identify the most influential set of source nodes based on their contribution to improving performance on target nodes. The latter further focuses more on the core nodes in the selected influential set, which closely align with the target nodes especially those presenting challenging predictions. Extensive experiments validate the efficacy of GraphInflu by overcoming the current state-of-the-art methods. Our code is available at https://github.com/lvXiangwei/GraphInflu.git.

Knowledge tracing (KT) is a fundamental task in intelligent education aimed at tracking students’ knowledge status and predicting their performance on new questions. The primary challenge in KT is accurately inferring a high-quality representation of students’ knowledge state that effectively captures their understanding of questions. However, existing methods are typically developed under the assumption that students’ behaviors directly reflect their knowledge state, which may not hold true especially in online learning scenarios. Abnormal behaviors exhibited by students, such as guessing and plagiarism, can introduce biases into the data, making it difficult to accurately assess students’ true knowledge state. To address this limitation, we propose a novel DebiAsed Cognition rEpresentation (DACE) modeling approach. This approach introduces a novel adversarial training strategy based on information bottleneck theory to obtain a debiased knowledge state representation that retains only the most reliable information for accurately predicting students’ performance on new questions. Moreover, we design a novel contrastive learning module through embedding-based augmentation to further enhance the robustness and generalizability of the learned knowledge state representation. We conduct extensive experiments on three public KT datasets and the newly released dataset BaiPy to demonstrate the superiority of our model over strong baselines, particularly when confronted with biased data. Our code and datasets are available at https://github.com/lvXiangwei/DACE.git .

Abstract argumentation is an important research area in AI. It is mainly about the acceptability of arguments in an argumentation framework. The classical notion of defense has not fully reflected some useful information implicitly encoded by the interaction relation between arguments. In this paper, instead of using arguments and attacks as first citizens, a novel notion of attack-defense is adopted as a first citizen, based on which a theory of attack-defense framework and attack-defense semantics are established, where an attack-defense is a triple (x,y,z), meaning that: an argument x defends an argument z against an attacker y. Attack-defense semantics can be used not only to identify the impact of arguments in some odd cycles, and remove some “useless” defenses, but also to capture new types of equivalence that cannot be represented by the existing notions of equivalence of argumentation frameworks. In addition, it shows that an attack-defense framework and attack-defense semantics can represent some knowledge that cannot be represented in Dung-style argumentation, e.g., some context-sensitive knowledge in a dialogue.

An autonomous system is constructed by a manufacturer, operates in a society subject to norms and laws, and interacts with end users. All of these actors are stakeholders affected by the behavior of the autonomous system. We address the challenge of how the ethical views of such stakeholders can be integrated in the behavior of an autonomous system. We propose an ethical recommendation component called Jiminy which uses techniques from normative systems and formal argumentation to reach moral agreements among stakeholders. A Jiminy represents the ethical views of each stakeholder by using normative systems, and has three ways of resolving moral dilemmas that involve the opinions of the stakeholders. First, the Jiminy considers how the arguments of the stakeholders relate to one another, which may already resolve the dilemma. Secondly, the Jiminy combines the normative systems of the stakeholders such that the combined expertise of the stakeholders may resolve the dilemma. Thirdly, and only if these two other methods have failed, the Jiminy uses context-sensitive rules to decide which of the stakeholders take preference over the others. At the abstract level, these three methods are characterized by adding arguments, adding attacks between arguments, and revising attacks between arguments. We show how a Jiminy can be used not only for ethical reasoning and collaborative decision-making, but also to provide explanations about ethical behavior.

Abstract In the field of collective argumentation, multiple agents may have different knowledge representations and individual preferences. In order to obtain a reasonable collective outcome for the group, either individual frameworks should be merged or individual preferences should be aggregated. However, framework merging and preference aggregation are different procedures, leading to disagreements on collective outcomes. In this paper, we figure out a solution to combine framework merging, argumentative reasoning and incomplete preference aggregation together. Furthermore, a couple of rational postulates are proposed to be the criteria for the rationality of collective outcomes obtained based on our approach.

Autonomous agents are supposed to be able to finish tasks or achieve goals that are assigned by their users through performing a sequence of actions. Since there might exist multiple plans that an agent can follow and each plan might promote or demote different values along each action, the agent should be able to resolve the conflicts between them and evaluate which plan he should follow. In this paper, we develop a logic-based framework that combines modal logic and argumentation for value-based practical reasoning with plans. Modal logic is used as a technique to represent and verify whether a plan with its local properties of value promotion or demotion can be followed to achieve an agent’s goal. We then propose an argumentation-based approach that allows an agent to reason about his plans in the form of supporting or objecting to a plan using the verification results.

New perspectives keep emerging in the logical study of social interactions, witnessed by yet another collection of research papers. Once our focus of reasoning shifts from an individual to a social setting, interesting issues naturally arise. The central question is the following: how is an individual’s attitude related to that of others and that of a group, especially when we take into account the informative communication between agents, as well as the structures of a group? This has been studied to various extent in social epistemology, social choice theory, social norms theory, etc. In this context, our main concern is logic: how can a logical approach shed light on our reasoning about these scenarios? We are not giving a comprehensive overview but just to mention some research that is mostly relevant. Communication between agents, public or private, has been extensively studied in dynamic epistemic logic (e.g. [1, 9, 10]) and dynamic epistemic logic has become a standard methodology in modelling dynamical changes of agent’s attitudes. Social network logics have been developed recently, adding social structures of a group as a core component to logic models and formal languages [2, 5–8, 11]. The papers included in this special issue provide us with further new insights about our understanding of social settings. In what follows, we briefly summarize their ideas and some main results.

This paper studies explanation semantics of argumentation by using a principle-based approach. In particular, we introduce and study explanation semantics associating with each accepted argument a set of such explanation arguments. We introduce various principles for explanation semantics for abstract argumentation, and list various relations among them. Then, we introduce explanation semantics based on defence graphs, and show which principles they satisfy.

An autonomous system is constructed by a manufacturer, operates in a society subject to norms and laws, and is interacting with end-users. We address the challenge of how the moral values and views of all stakeholders can be integrated and reflected in the moral behavior of the autonomous system. We propose an artificial moral agent architecture that uses techniques from normative systems and formal argumentation to reach moral agreements among stakeholders. We show how our architecture can be used not only for ethical practical reasoning and collaborative decision-making, but also for the explanation of such moral behavior.

To resolve conflicts amongst norms, various non-monotonic formalisms can be used to perform prioritized normative reasoning. Meanwhile, formal argumentation provides a way to represent non-monotonic logics. In this paper we propose a representation of prioritized normative reasoning by argumentation. Using hierarchical abstract normative systems (HANS), we define three kinds of prioritized normative reasoning approaches called Greedy, Reduction and Optimization. Then, after formulating an argumentation theory for a HANS, we show that for a totally ordered HANS, Greedy and Reduction can be represented in argumentation by applying the weakest link and the last link principles, respectively, and Optimization can be represented by introducing additional defeats capturing the idea that for each argument that contains a norm not belonging to the maximal obeyable set then this argument should be rejected.

In Dung's abstract argumentation theory, an extension can be represented by subsets of it in the sense that from each of these subsets, the extension can be obtained again by iteratively applying the characteristic function. Such so-called regular representations can be used to differentiate argumentation frameworks having the same extensions. In this paper we provide a full characterization of relations between seven different types of representation equivalence.

The existing approaches to formulate the semantics of probabilistic argumentation are based on the notion of possible world. Given a probabilistic argument graph (PrAG) with |$n$| nodes, up to |$2^n$| subgraphs are blindly constructed and their extensions under a given semantics are computed. Then, the probability of a set of arguments |$E$| being an extension under a given semantics |$\sigma$| (denoted as |$p(E^\sigma)$|⁠) is equal to the sum of the probabilities of all subgraphs each of which has the extension |$E$|⁠. Since many irrelevant subgraphs are constructed, and in many cases, computing extensions of subgraphs is computationally intractable, these approaches are fundamentally inefficient or infeasible. In existing literature, while approximate approaches based on the Monte Carlo simulation technique have been proposed to estimate the probability of extensions, how to improve the efficiency of computation without using the simulation technique is still an open problem. In this article, we address this problem from the following two perspectives. First, conceptually, we define specific properties to characterize the subgraphs of a PrAG with respect to a given extension, such that the probability of a set of arguments |$E$| being an extension can be defined in terms of these properties, without (or with less) construction of subgraphs. Second, computationally, we take preferred semantics as an example, and develop algorithms to evaluate the efficiency of our approach. The results show that our approach not only dramatically decreases the time for computing |$p(E^\sigma)$|⁠, but also has an attractive property, which is contrary to that of existing approaches: the denser the edges of a PrAG are or the bigger the size of a given extension |$E$| is, the more efficient our approach computes |$p(E^\sigma)$|⁠. Meanwhile, it is shown that under complete and preferred semantics, the problems of determining |$p(E^\sigma)$| are fixed-parameter tractable.

The interpretation of a norm is often uncertain and conflicting. In this paper we propose a model for arguing about legal interpretation, which considers the problems of vagueness. After motivating our adoption of graded categories as a tool to tackle the problem of open texture in legal interpretation, we introduce a model based on fuzzy logic and argumentation. Then, we conduct a case study by using an example from medically assisted reproduction.

In various argumentation systems, under most of situations, only the status of some arguments of the systems should be evaluated, while that of others is not necessary to be figured out. Based on this observation, we first introduce an efficient method to evaluate the status of a part of arguments in an abstract argumentation framework (AF). Given an AF and a subset of arguments within it, the minimal set of arguments that are relevant to this subset (called the set of relevant arguments of the subset) is identified. Under a semantics satisfying the directionality criterion, the set of extensions of the sub-framework induced by the set of relevant arguments of the given subset (called a partial semantics of the AF with respect to this subset) can be evaluated locally. Then, we introduce three basic properties of the partial semantics of argumentation: monotonicity, extensibility and combinability, which lay a foundation for developing efficient algorithms for the status evaluation of a part of arguments in an AF. Finally, we conduct an empirical investigation on the properties of computing the partial semantics of argumentation using answer-set programming. The average results show that the computational benefits of this method is closely related to the edge density of the defeat graph of a given AF.

Currently, only some argumentation frameworks (AFs) with special topologies have been identified as tractable classes. By taking advantage of the tractability of some parts of a general AF, this paper proposes a method to compute the extensions of an AF based on its strongly connected components (SCCs). In this method, an AF is partitioned into a set of sub frameworks (by a linear time algorithm) according to its SCCs. Under an argumentation semantics that satisfies the criterion of directionality, sets of extensions of all sub-frameworks can be computed locally, and combined incrementally to form the extensions of the original AF. The analysis shows that by using this method the complexity of computing the extensions of an AF may be reduced to a greater or a lesser extent, depending on the size and the topology of a dominant sub-framework of the AF, as well as the number of extensions of the AF.

Currently, little attention has been paid to the agent architecture that allows rules (including non-ground rules) to be inserted into or deleted from the agents' knowledge base at run-time. Meanwhile, it is unclear that how the theories for reasoning and decision-making are generated and updated in accordance with a set of changing rules, and how agents deliberate according to the dynamic theories. In this paper, we introduce a flexible agent based on the language of defeasible logic and the formalism of argumentation. The architecture of flexible agent is similar to BDI (beliefs-desires-intentions), but the knowledge for decision-making (a set of non-ground rules) is allowed to be changed at run-time, and the reasoning process is non-monotonic. After presenting the architecture of flexible agent, we show how the theories for agent reasoning are represented, generated and updated. Then, an argumentation-based deliberation method with dynamic theories is put forward. Different from the process of argument construction in the existing argumentation theories, the one in flexible agent is non-monotonic, i.e., the arguments and the defeat relations among them can be updated (added or subtracted) with the variation of theories. We show that this novel agent is both autonomous and flexible, so it is adaptable to the open and dynamic environment where the observations and the external business requirements are changeable.

Although EDI (electronic data interchange) have provided a useful method to the automated message interchanging between enterprises, this method costs a lot of money and resources, there is a new technology named ebXML (electronic business using eXtensible Markup Language) BPSS (business process specification schema) which provides a more easy way to integrate the whole business process. EbXML BPSS supports the specification of the set of elements required to configure a runtime system in order to execute a set of ebXML business transactions. EbXML BPSS describes which message should be interchanged and how to interchange, and provides a set of specification for the process automation. In this study, we propose a federated multi-agents system, which can implement the business process, and describe a detailed mechanism to apply the BPS to a prototype implementation.