---
title: CONCAP in Claim-Based Argumentation
url: https://www.emergentmind.com/topics/concap
type: topic
---

# CONCAP in Claim-Based Argumentation

Searching arXiv for the specified paper and closely related work on claim-based semantics and concurrence in CAFs.
CONCAP denotes the concurrence problem for claim-based semantics in well-formed claim-based argumentation frameworks (CAFs) under naive semantics. In this setting, concurrence means that the inherited claim-based semantics and the claim-level semantics yield exactly the same claim-based extensions. The note “Concurrence for well-formed CAFs: Naive Semantics” establishes that deciding this property remains computationally hard even under well-formedness, showing that the problem is coNP-hard and, together with the inherited upper bound, coNP-complete [2112.10445].

## 1. Formal setting

An abstract argumentation framework (AF) is a directed graph $F = (A, R)$ with $A \subseteq U$ and $R \subseteq A \times A$, where $U$ is a fixed universe of arguments. For $a \in A$, the set of outgoing attacks is denoted by
$$
a_F^+ := \{ b \in A \mid (a, b) \in R \}.
$$
A set $E \subseteq A$ is conflict-free in $F$ iff there are no $a, b \in E$ with $(a, b) \in R$. Writing $cf(F)$ for the family of conflict-free sets, naive semantics returns the $\subseteq$-maximal conflict-free sets:
$$
E \in na(F) \iff E \in cf(F) \text{ and there is no } E' \in cf(F) \text{ with } E \subset E'.
$$

A claim-augmented argumentation framework is a triple $F = (A, R, cl)$, where $(A, R)$ is an AF and $cl: A \to C$ maps each argument to a claim in a claim set $C$. This map extends to sets by
$$
cl(E) := \{ cl(a) \mid a \in E \}
$$
for $E \subseteq A$ [2112.10445].

A CAF $F = (A, R, cl)$ is well-formed iff arguments with the same claim have identical sets of attacked arguments:
$$
\forall a, b \in A,\; \text{if } cl(a) = cl(b), \text{ then } a_F^+ = b_F^+.
$$
Equivalently, all arguments labeled by the same claim are indistinguishable with respect to outgoing attacks. This restriction is central to CONCAP because it enforces claim-coherent attack behavior without collapsing the framework to the claim level [2112.10445].

## 2. Claim-based naive semantics and concurrence

For an AF semantics $\sigma$, the inherited claim-based semantics is
$$
\sigma_c(F) := \{ cl(E) \mid E \in \sigma(F) \}.
$$
Specializing to naive semantics, the inherited claim-based extensions are
$$
Ext_i(F) := na_c(F) := \{ cl(E) \mid E \in na(F) \}.
$$

Claim-level naive semantics is defined through an induced claim-level attack relation $R_c$ on claims:
$$
(p, q) \in R_c \iff \exists a, b \in A \text{ such that } cl(a) = p,\; cl(b) = q,\; \text{and } (a, b) \in R.
$$
A set of claims $S \subseteq cl(A)$ is claim-conflict-free iff no $p, q \in S$ have $(p, q) \in R_c$. Writing $cf_c(F)$ for all claim-conflict-free sets, the claim-level naive semantics is
$$
Ext_{cl}(F) := na_{cl}(F) := \{ S \subseteq cl(A) \mid S \in cf_c(F) \text{ and there is no } S' \in cf_c(F) \text{ with } S \subset S' \}.
$$

It is known that $Ext_{cl}(F) \subseteq Ext_i(F)$. Concurrence under naive semantics is therefore the condition
$$
Ext_i(F) = Ext_{cl}(F).
$$
For well-formed CAFs, CONCAP asks whether this equality holds. Formally, the decision problem is
$$
Con_{na}^{wf} := \{ F \mid F \text{ is a well-formed CAF and } Ext_i(F) = Ext_{cl}(F) \}.
$$

A central characterization reused in the note states that, for every CAF $F$,
$$
Ext_i(F) = Ext_{cl}(F)
$$
if and only if $Ext_i(F)$ is incomparable, meaning that for any distinct $X, Y \in Ext_i(F)$, neither $X \subset Y$ nor $Y \subset X$ holds [2112.10445]. In effect, concurrence under naive semantics reduces to an antichain condition on inherited claim sets.

## 3. Decision problem and complexity classification

The paper studies the problem:

> Given a well-formed CAF $F = (A, R, cl)$, decide whether $Ext_i(F) = Ext_{cl}(F)$.

By the incomparability characterization, this is equivalent to deciding whether $Ext_i(F)$ is incomparable. The note proves that $Con_{na}^{wf}$ is coNP-complete [2112.10445].

The upper bound follows from the known complexity of concurrence for general CAFs under naive semantics. Since well-formed CAFs form a subclass, membership in coNP is inherited. Concretely, the complement can be decided in NP by guessing two argument-level naive extensions $E, G \in na(F)$ and checking whether $cl(E) \subset cl(G)$. Verifying $E \in na(F)$ is polynomial: first check conflict-freeness, and then for each $a \in A \setminus E$ verify that $E \cup \{a\}$ is not conflict-free. This gives an NP witness for non-concurrence and places the complement in NP [2112.10445].

The lower bound is the substantive contribution of the note. It gives a many-one reduction from UNSAT to $Con_{na}^{wf}$, establishing coNP-hardness even when the input is restricted to well-formed CAFs. Together with the upper bound, this yields coNP-completeness. The result resolves an open problem left by Dvořák et al. (2021), namely whether well-formedness reduces the complexity of concurrence under naive semantics [2112.10445].

## 4. Reduction from UNSAT

The source problem is UNSAT, the set of propositional CNF formulas that are not satisfiable. Given a CNF formula $\varphi$ over variables $X$ with clauses $C = \{cl_1, \dots, cl_n\}$, assuming that $\varphi$ has no tautological clause, the reduction constructs a well-formed CAF $F = (A, R, cl)$ as follows:
- Arguments:
$$
A = X \cup \bar{X} \cup C \cup \{ \varphi \} \cup \{ a_1, a_2 \},
$$
where $\bar{X} = \{ \bar{x} \mid x \in X \}$.
- Attack relation:
$$
R =
\{ (x, cl) \mid cl \in C \text{ and } x \in cl \} \cup
\{ (\bar{x}, cl) \mid cl \in C \text{ and } \neg x \in cl \}
$$
$$
\cup\; \{ (x, \bar{x}), (\bar{x}, x) \mid x \in X \}
\cup\; \{ (cl_i, \varphi) \mid i \in \{1, \dots, n\} \}
\cup\; \{ (\varphi, a_2) \}.
$$
- Claim labeling:
$$
cl(x) = x,\quad cl(\bar{x}) = \bar{x},\quad cl(cl_i) = cl_i,\quad cl(\varphi) = \varphi,\quad cl(a_1) = cl(a_2) = a.
$$

Well-formedness holds because the only pair of arguments with the same claim is $a_1$ and $a_2$, and both have no outgoing attacks, so
$$
a_{1\,F}^+ = a_{2\,F}^+ = \emptyset.
$$
All other arguments have distinct claims, so the well-formedness condition is vacuous for them [2112.10445].

The reduction is driven by a small gadget $\{a_1, a_2, \varphi\}$ with $\varphi \to a_2$. At the argument level, this gadget can generate two incompatible choices: include $\varphi$ and exclude $a_2$, or exclude $\varphi$ and include $a_2$. The clause–literal part of the construction ensures that excluding $\varphi$ without introducing clauses is possible iff $\varphi$ is satisfiable. This mechanism creates, in the satisfiable case, two naive extensions whose claim sets are strictly ordered by inclusion, which violates incomparability and therefore concurrence [2112.10445].

## 5. Satisfiable and unsatisfiable cases

If $\varphi$ is satisfiable with a model $M \subseteq X$, define
$$
E := M \cup \{ \bar{x} \mid x \notin M \} \cup \{ \varphi, a_1 \}
$$
and
$$
E' := M \cup \{ \bar{x} \mid x \notin M \} \cup \{ a_1, a_2 \}.
$$
Then $E, E' \in na(F)$, and their claim sets satisfy
$$
cl(E) = M \cup \{ \bar{x} \mid x \notin M \} \cup \{ \varphi, a \},
$$
$$
cl(E') = M \cup \{ \bar{x} \mid x \notin M \} \cup \{ a \},
$$
so that
$$
cl(E') \subset cl(E).
$$
Hence $Ext_i(F)$ is not incomparable, implying $Ext_i(F) \ne Ext_{cl}(F)$; the framework is not concurrent [2112.10445].

The proof sketch is direct. Both sets are conflict-free because each variable contributes exactly one of $x$ or $\bar{x}$, avoiding the mutual attacks between literals, and every clause is attacked by some literal selected according to $M$. Moreover, $a_2$ cannot be added to $E$ because it conflicts with $\varphi$, while $\varphi$ cannot be added to $E'$ because it conflicts with $a_2$. This establishes maximality among conflict-free sets.

If $\varphi$ is unsatisfiable, then $Ext_i(F)$ is incomparable, and therefore
$$
Ext_i(F) = Ext_{cl}(F).
$$
The paper’s proof sketch distinguishes extensions containing $\varphi$ from those not containing it. If $\varphi \in E$, then no clause argument can be in $E$, because each clause attacks $\varphi$, and maximality forces $E$ to contain exactly one of $x$ or $\bar{x}$ for each variable and also $a_1$. If $\varphi \notin E$, then $E$ must contain a clause $cl_i$; otherwise $\varphi$ could be added. In that case, the presence of a clause claim prevents strict inclusion with extensions containing $\varphi$. Altogether, all claim sets in $Ext_i(F)$ are pairwise incomparable [2112.10445].

Combining the two directions yields the theorem:
$$
\varphi \in UNSAT \iff F \in Con_{na}^{wf}.
$$
This establishes coNP-hardness by polynomial reduction from UNSAT.

## 6. Example, significance, and boundaries

The note includes an illustrative CNF formula with variables $x_1, x_2, x_3, x_4$ and clauses
$$
cl_1 = \{ x_1, x_3, x_4 \},\quad
cl_2 = \{ \neg x_3, \neg x_4, \neg x_2 \},\quad
cl_3 = \{ \neg x_1, \neg x_3, x_2 \}.
$$
Building the CAF by the reduction and taking the model $M = \{x_1, x_2, x_3\}$ with $x_4$ false yields two naive extensions:
$$
E = \{ x_1, x_2, x_3, \bar{x}_4, \varphi, a_1 \},
$$
$$
E' = \{ x_1, x_2, x_3, \bar{x}_4, a_1, a_2 \}.
$$
Their claim sets are
$$
cl(E) = \{ x_1, x_2, x_3, \bar{x}_4, \varphi, a \},
$$
$$
cl(E') = \{ x_1, x_2, x_3, \bar{x}_4, a \},
$$
hence $cl(E') \subset cl(E)$. This witnesses non-concurrence in the satisfiable case [2112.10445].

The significance of the result is primarily negative but precise. Well-formedness is a natural restriction: it guarantees that arguments sharing a claim have identical outgoing attacks. One might expect this to simplify reasoning about concurrence. The paper shows that it does not. The complexity remains coNP-complete, matching the status already known for general CAFs under naive semantics [2112.10445].

The note does not identify any tractable subclasses of well-formed CAFs for which CONCAP under naive semantics becomes polynomial. It also does not address other semantics such as grounded, preferred, or stable for well-formed CAFs. A plausible implication is that further tractability results, if they exist, must rely on additional structural restrictions beyond well-formedness alone.

## 7. Algorithmic implications and open directions

From an algorithmic perspective, the coNP characterization immediately yields an NP procedure for the complement: guess $E, G \in na(F)$ and accept if $cl(E) \subset cl(G)$. This suggests practical SAT or ILP encodings for searching for two naive extensions whose claim sets are strictly ordered. By contrast, explicitly computing all of $Ext_i(F)$ and then checking incomparability may be exponential because the number of naive extensions can be exponential [2112.10445].

The note also discusses a possible role for heuristics. One can search for “claim witnesses” whose difference resembles the reduction gadget, or exploit well-formedness to compress a framework by merging arguments with identical claim and out-neighborhood. This suggests implementation strategies, but the paper does not present a tractability theorem or a complete algorithmic framework for such cases [2112.10445].

In claim-based reasoning more broadly, concurrence is attractive because it guarantees consistency across claim-based lifting variants: inherited and claim-level selections coincide. This removes one source of semantic ambiguity in tool design. The complexity result shows, however, that verifying this agreement is computationally intractable in the worst case, even under a natural coherence restriction. The paper therefore settles a specific open question while leaving the broader boundary between tractable and intractable cases unresolved [2112.10445].

Source: https://www.emergentmind.com/topics/concap