Causal Discovery as Dialectical Aggregation: A Quantitative Argumentation Framework

TL;DR

QACD introduces a semantics-driven, argumentation-based framework for causal discovery that improves robustness and accuracy in noisy, finite-sample scenarios by aggregating conflicting evidence.

Contribution

It presents QACD, a novel approach that models CI test outcomes as graded arguments and uses fixed-point semantics for more reliable causal structure learning.

Findings

QACD enhances structural coherence in noisy regimes.

QACD outperforms classical methods in interventional reliability.

QACD remains competitive with existing baselines.

Abstract

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.