Back to AI News
arxiv

CaDO: Formal Verification-Guided Multi-Agent Framework for Causal Dataset Repair and Generation

Researchers present CaDO, a multi-agent framework for automatically diagnosing and repairing causal reasoning datasets. It formalizes natural-language causal questions, uses critic agents for sample-level and dataset-level diagnosis, and applies reusable repair operators plus explorer and expander agents to correct and generate data. CaDO outperformed the strongest of five data-cleaning baselines in 16 of 18 settings and improved downstream model accuracy by up to 9.1%; at least 95% of flagged samples were defective. On CLadder, it incorrectly modified 92.2% fewer correct labels than an LLM-based judge.

AI Brief

It offers a concrete, code-backed method for improving causal reasoning datasets and provides quantitative evidence of cleaner labels and better downstream performance.

arXiv:2611.24406v1 Announce Type: new Abstract: High-quality causal reasoning datasets are essential for developing models with reliable causal understanding, particularly in specialized domains. However, existing datasets frequently suffer from incorrect annotations and structural imbalances, which can lead to biased predictions and limited generalization. To address this issue, we propose CaDO (Causal-Engine-Guided Multi-Agent Data Optimization), a multi-agent framework for automated diagnosis and repair of causal reasoning datasets. Our framework consists of three key stages: formal causal reasoning, multi-level diagnosis, and automated data repair and generation. First, a Formalizer Agent translates natural language questions into formal causal queries to minimize semantic deviations and make retrieval and verification more explicit and tractable. Second, a dual-level diagnosis mechanism employs independent Critic Agents to identify sample-level errors while analyzing query-type coverage and label distributions to detect dataset-level structural imbalances. Finally, the framework employs reusable repair operators grounded in formal causal semantics to correct erroneous samples, while Explorer and Expander Agent collaboratively produce diverse and semantically consistent new samples. Experimental results show that CaDO significantly outperforms the strongest of five data cleaning baselines in 16 out of 18 experimental settings, improving downstream model accuracy by up to 9.1%. Moreover, at least 95% of the samples flagged by CaDO are defective, and on CLadder, CaDO incorrectly modifies 92.2% fewer correct labels than an LLM-based judge, demonstrating its effectiveness and reliability. Our code and data are publicly available at https://github.com/thisissyu/cado.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!