AI
Optimal Correction Sets for Provable Redress of Contestable Causal Discovery
Publication: cs.AI updates on arXiv.org
Author: Fabrizio Russo, Giuseppe Mazzotta, Carmine Dodaro, Francesco Ricca, Francesca Toni
Source published: 27/10/2026, 04:00:00 UTC
Briefing created: 22/11/2026, 16:51:57 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 28/10/2026, 02:33:24 UTC.
Briefings
This article proposes a formal framework for contestable causal discovery using Constraint-based Causal Discovery (CBD) and Causal Assumption-based Argumentation (ABA). The authors argue that existing repair procedures lack formal justification for user contestations, proposing instead a minimum-cost correction problem solved via Answer Set Programming (ASP). The study establishes a two-way correspondence between optimal stable models and minimum-cost correction sets, ensuring fidelity to retained facts, subset-minimality, and cost optimality. An empirical evaluation on common benchmarks shows that the proposed OptABA-PC method retains more correct facts, reduces ambiguity, and improves graph reconstruction compared to Causal ABA.
Key points
- The study formalizes contestable causal discovery as a minimum-cost correction problem solvable via Answer Set Programming (ASP).
- OptABA-PC achieves better graph reconstruction and reduced ambiguity on benchmarks compared to Causal ABA.
Context
Constraint-based causal discovery relies on conditional-independence tests that may conflict in finite samples, leaving no causal graph satisfying all tests. Causal Assumption-based Argumentation (ABA) makes these tests explicit as defeasible commitments but lacks formal response to user contestations. This work addresses this gap by proposing a formal repair object and justification system.
Limitations
The empirical evaluation is limited to common benchmarks and does not include a full-scale dataset. The study relies on the availability of the arXiv preprint, which may not reflect final publication details or complete results.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 22/11/2026, 16:51:57 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeiggm2s3fnnjttiaylqcmu5icaoorenxvtgaqtwlgxlnrgovc6nzai