CArLA

Contestability & Redress

LINDA 2026 ORAL

Optimal Correction Sets for Argumentative Causal Discovery

Recasts the release of inconsistent causal constraints as a cost-aware repair problem using optimal minimal correction sets.

The problem

Noisy conditional-independence tests can be mutually incompatible. A repair step must decide which facts to release without hiding that decision.

The approach

The work transfers minimal correction set reasoning from Answer Set Programming into Causal ABA, defines optimal correction sets using release costs, and computes them through standard weak constraints.

What the work establishes

  • Substantially increases the identification rate of true constraints in the reported experiments.
  • Reduces the number of compatible output causal graphs.
  • Improves graph reconstruction according to standard metrics, with an explicit optimisation/runtime trade-off.

Why it matters to CArLA

Optimal correction sets make inconsistency handling principled and auditable: CArLA can explain which constraints were released, why they were selected, and which conclusions remain supported.

Citation

Russo, F., Mazzotta, G., Dodaro, C., Ricca, F., & Toni, F. (2026). Optimal Correction Sets for Argumentative Causal Discovery. LINDA 2026, Lisbon, Portugal.