Formal argumentation provides a mathematical scaffold for reasoning, where arguments are nodes in a graph and attacks are directed edges. To determine which arguments "win," researchers apply semantics—sets of rules that define acceptable extensions of arguments. While the field has established a core set of semantics (such as stable, preferred, grounded, and complete), a persistent challenge has emerged: as researchers develop more sophisticated reasoning frameworks—handling uncertainty, dynamics, or fuzzy logic—their tools and theoretical results often remain restricted to a narrow subset of these rules. This "semantic coverage gap" limits the generalizability of modern argumentation systems.
The Core vs. The Nuanced
The foundation of the field, established by Dung, focuses on four primary semantics. However, practical applications often require more nuanced interpretations. For instance, semi-stable and stage semantics are useful for ensuring that an extension is as large as possible even when no stable extension exists. Ideal and eager semantics provide a unique, skeptical middle ground between the grounded and preferred approaches.
Despite their utility, these nuanced semantics are frequently left out of new theoretical advancements. Gianvincenzo Alfano et al. note this in the context of General Epistemic Abstract Argumentation Frameworks, where core semantics are well-defined but semi-stable or ideal interpretations remain an ongoing challenge. Similarly, Johannes Klaus Fichte et al. highlight that while facet-based reasoning offers a new way to measure argument significance, a complete theoretical treatment for stage and semi-stable semantics is still missing.
The Computational Bottleneck
The gap is most visible in the development of high-performance solvers. Modern reasoning tasks are often reduced to Boolean satisfiability (SAT) or Answer-Set Programming (ASP) problems. However, a reduction that works for one semantic type rarely translates directly to another.
- ASP Encodings: Norbert Manthey et al. developed efficient ASP encodings using conditional literals, but these have not yet been extended to ideal or eager extensions. Similarly, Gerhard Brewka et al. demonstrate that dynamic encoding in ASP can solve complex tasks at the third level of the polynomial hierarchy, yet adapting this to grounded or ideal semantics remains future work.
- SAT Solvers: Federico Cerutti et al. introduced PrefSat, which uses iterative SAT calls to enumerate preferred extensions. However, the authors point out that supporting alternative semantics like semi-stable or stage requires deriving entirely new sets of non-redundant logical constraints.
Reasoning Under Uncertainty and Change
When argumentation frameworks become dynamic or incomplete, the semantic gap complicates the landscape of complexity and decidability.
In Incomplete Argumentation Frameworks (iAFs), where the existence of arguments or attacks is uncertain, Andreas Niskanen et al. provided the first SAT-based algorithms for argument acceptance, but these were restricted to conflict-free, admissible, and stable semantics. Dorothea Baumeister et al. further mapped the complexity of verification in these uncertain environments, noting that the framework still needs to be extended to non-classical semantics to provide a comprehensive theoretical picture.
The same pattern appears in Belief Revision and Enforcement. Martin Diller et al. found that frameworks for updating knowledge are often restricted to "I-maximal" semantics (where no extension is a subset of another), effectively excluding complete semantics. When modifying a framework to force an argument's acceptance, Adrian Haret et al. established a duality between revision and enforcement, but primarily for admissible semantics.
Bridging the Gap
The push for broader semantic coverage is not merely for the sake of completeness; it is necessary for the cross-pollination of argumentation with other fields. Alexander Bochman shows that causal reasoning can be unified with argumentation, but this currently relies heavily on stable semantics. Similarly, Francesca Toni explores how argumentation can justify logic programming outputs, yet these explanation techniques are currently limited to specific logic paradigms.
Closing this gap requires moving beyond "one-size-fits-all" algorithms. As Jesse Heyninck et al. demonstrate in Assumption-Based Argumentation, restoring critical properties like the existence of extensions in complex frameworks often requires developing entirely new algebraic structures, such as six-valued labeling systems. The future of the field lies in ensuring that these sophisticated reasoning tools can support the full spectrum of logical interpretations.
Key Foundational Work
Logic Programs with Abstract Constraint Atoms (2004) — contributed 1 concept elements to this direction.
Argumentation in Representation Semantics (1984) — contributed 1 concept elements to this direction.