Abstract: This paper addresses a subtle but significant technical flaw in $ASPIC^+$, one of the most popular frameworks for structured argumentation. By rethinking how conflicts are represented—moving from binary "argument vs. argument" attacks to "sets of arguments vs. argument" conflicts—the authors resolve a logic "collapse" that occurs when dealing with complex, non-binary contradictions. The proposed formalism introduces an extended contrariness relation and a set-based argumentation framework that preserves logical soundness and expressiveness.
Introduction: The Expressiveness Collapse
In formal argumentation, the $ASPIC^+$ framework serves as a bridge between high-level abstract graphs and low-level logic. A defining feature of $ASPIC^+$ is generic contrariness. Unlike simple binary negation (e.g., "It is raining" vs. "It is not raining"), generic contrariness allows for more nuanced, multi-way conflicts. For instance, the status of being a "Bachelor" is contrary to both being "Married" and being "Divorced."
However, a technical flaw arises when combining this generic contrariness with standard logical closure rules, such as transposition (if $A$ implies $B$, then "not $B$" implies "not $A$"). Because "not $B$" can represent multiple distinct states in a generic system, a naive application of these rules leads to what the authors term a "logical collapse." In certain scenarios, the system becomes so restricted that it forces contradictions from empty sets of premises just to maintain internal consistency. Essentially, the current iteration of $ASPIC^+$ fails to handle situations where a single concept has multiple, mutually exclusive opposites.
Methodology: Redefining Conflict
To fix this collapse, the authors propose a two-pronged revision to the $ASPIC^+$ formalism:
1. Extended Contrariness Relation ($EC$)
Instead of defining contrariness only between individual formulas, the authors introduce a notion of closure for the contrariness relation itself. This involves identifying sets of formulas that are collectively incompatible based on the strict rules of the system. By moving the closure property to the contrariness relation rather than just the inference rules, they prevent the logical chain reactions that lead to collapse.
2. Relevant-Set-Based Argumentation Framework ($RS-AF$)
Standard argumentation frameworks, such as Dung’s AFs, rely on binary attacks where one argument attacks another. The authors shift this to a "relevant-set-based" approach. In this model, a set of arguments $\Sigma$ can collectively attack an argument $b$. This allows the system to explicitly represent "ternary" or higher-order conflicts where no single argument is the culprit, but the combination of arguments is impossible.
Key Insights: The "Married or 16?" Example
The "Married or 16?" example perfectly illustrates the flaw in the original framework. Consider a teenager ($T$) who is a bachelor ($B$). If being married ($M$) or divorced ($D$) are both contraries of being a bachelor, a standard system might incorrectly conclude that you cannot even propose that someone is married if you know they are a teenager. This leads to counterintuitive limitations on what can be expressed within the logic.
By using the $RS-AF$ approach, the system can identify that while $T$ and $B$ are compatible, and $T$ and $M$ might be compatible in some contexts, the specific combination of premises that leads to a violation of the contrariness relation is flagged as a collective conflict. This preserves the ability to model complex real-world contradictions without sacrificing logical integrity.
Results and Rationality Postulates
The authors mathematically prove that this new structure satisfies the four "rationality postulates" which are the gold standard for ensuring an argumentation system remains sound:
- Sub-argument Closure: If an argument is accepted, all its parts must be accepted.
- Closure under Strict Rules: If a set of arguments is accepted, any conclusion reached by applying strict rules to them must also be accepted.
- Direct Consistency: The set of accepted conclusions must not contain any direct contraries.
- Indirect Consistency: The closure of the accepted conclusions under strict rules must be consistent.
Unlike previous attempts to fix $ASPIC^+$, which often forced the language back into standard binary negation, this method preserves full expressiveness, allowing an arbitrary number of contraries for any given formula.
Conclusion and Future Work
By shifting the focus from individual arguments to collective sets of arguments, the authors provide a robust solution to the expressiveness collapse in structured argumentation. While this approach is more computationally complex—requiring the calculation of the closure of the contrariness relation—it ensures that $ASPIC^+$ can handle the nuances of real-world reasoning. Future work will focus on optimizing the computation of these set-based conflicts and exploring translations to other abstract argumentation systems.
Summary updated