[Paper Review] Probabilistic Argumentation. An Equational Approach

Michael Gabbay, Odinaldo Rodrigues|arXiv (Cornell University)|Mar 18, 2015

Multi-Agent Systems and Negotiation8 references5 citations

TL;DR

This paper introduces a novel equational approach to probabilistic argumentation by translating abstract argumentation frameworks into classical propositional logic, then importing probabilistic semantics via model-based probability assignments. The key contribution is a principled, logic-translation-based method that ensures probabilistic interpretations are semantically grounded in the logical structure of argumentation networks, offering a coherent alternative to ad hoc probability assignments in existing frameworks.

ABSTRACT

There is a generic way to add any new feature to a system. It involves 1) identifying the basic units which build up the system and 2) introducing the new feature to each of these basic units. In the case where the system is argumentation and the feature is probabilistic we have the following. The basic units are: a. the nature of the arguments involved; b. the membership relation in the set S of arguments; c. the attack relation; and d. the choice of extensions. Generically to add a new aspect (probabilistic, or fuzzy, or temporal, etc) to an argumentation network can be done by adding this feature to each component a-d. This is a brute-force method and may yield a non-intuitive or meaningful result. A better way is to meaningfully translate the object system into another target system which does have the aspect required and then let the target system endow the aspect on the initial system. In our case we translate argumentation into classical propositional logic and get probabilistic argumentation from the translation. Of course what we get depends on how we translate. In fact, in this paper we introduce probabilistic semantics to abstract argumentation theory based on the equational approach to argumentation networks. We then compare our semantics with existing proposals in the literature including the approaches by M. Thimm and by A. Hunter. Our methodology in general is discussed in the conclusion.

Motivation & Objective

• To develop a principled, semantics-grounded method for assigning probabilities to abstract argumentation frameworks.
• To address the limitations of ad hoc probability assignments in existing probabilistic argumentation systems.
• To establish a systematic methodology—'Logic by Translation'—that ensures probabilistic semantics align with the dynamic and structural features of argumentation networks.
• To compare the proposed equational approach with existing frameworks, such as those by Thimm and Hunter, highlighting conceptual and technical distinctions.

Proposed method

• Translate an abstract argumentation framework ⟨S, R⟩ into a classical propositional logic theory Δ⟨S,R⟩ by treating arguments in S as atomic propositions and using R to generate logical equations.
• Express attack relations via equational constraints: x ↔ ⋀ᵢ ¬yᵢ, which are reformulated in real-valued semantics as x = 1 − max{yᵢ} for probabilistic interpretation.
• Interpret the equational system over the unit interval [0,1], with values 0 (out), 1 (in), and 1/2 (undetermined), mapping to Kleene’s three-valued logic.
• Assign probabilities to classical models of the propositional theory Δ⟨S,R⟩, then export these probabilities back to the argumentation framework via the equational constraints.
• Use the equational system (E3) to ensure consistency between argument acceptability and probability assignments, distinguishing the approach from weaker probabilistic definitions.
• Compare the proposed method with Thimm’s and Hunter’s approaches, emphasizing the conceptual necessity of strong equations (E3) for coherent semantics.

Experimental results

Research questions

• RQ1 How can probabilistic semantics be meaningfully assigned to abstract argumentation frameworks without distorting their dynamic and structural properties?
• RQ2 What is the role of logical translation in grounding probabilistic interpretations in argumentation systems?
• RQ3 How does the proposed equational approach differ conceptually and technically from existing probabilistic argumentation frameworks such as Thimm’s and Hunter’s?
• RQ4 Can a logic-translation methodology ensure that probability assignments are both coherent and semantically aligned with argumentation semantics?
• RQ5 What is the significance of the strong equation (E3) in ensuring consistency between argument acceptability and probabilistic models?

Key findings

• The equational approach provides a coherent, semantics-grounded method for probabilistic argumentation by translating argumentation networks into classical propositional logic and assigning probabilities to models.
• The method ensures that probabilistic assignments are consistent with the logical structure of attacks, using equations like x = 1 − max{yᵢ} to define acceptability in terms of attacker states.
• The approach differs fundamentally from Thimm’s by requiring the strong equation (E3), which enforces consistency between argument status and probability, whereas Thimm’s approach allows inconsistent values for symmetric nodes.
• The method supports both internal (model-based) and external (extension-based) views of probability, with approximation results showing convergence under certain conditions.
• The framework is extensible to other features like fuzziness or temporality by applying the same logic-translation methodology to classical logic with those features.
• The authors demonstrate that their approach avoids technical compromises found in other systems, such as Thimm’s reliance on weaker inequalities, by grounding probability in model-theoretic semantics.