Causal Ensemble Agent: Hierarchical Causal Discovery with LLM-guided Expert Reweighting

2026-06-09

AI Summary

This study addresses the inconsistency in causal discovery results arising from algorithmic differences and the frequent neglect of domain knowledge, which hinder the construction of accurate causal graphs. To overcome these limitations, the authors propose a linear opinion pool framework that integrates multiple hierarchical statistical causal discovery experts and, for the first time, employs a large language model (LLM) as a meta-judge. The LLM dynamically reweights expert opinions when aggregated confidence approaches the decision boundary, thereby synergistically combining data-driven evidence with semantic information to construct hierarchical causal graphs. Empirical evaluations on both synthetic and real-world datasets demonstrate that the proposed method significantly outperforms existing approaches, yielding more complete and reliable causal structures.

Abstract

Causal discovery aims to uncover causal structures from observational data, which is crucial for real-world decision-making. However, different causal discovery algorithms can produce divergent results that conflict with each other, complicating the identification of accurate causal graphs. Traditional approaches rely on numerical values and statistical assumptions, often ignoring rich domain-specific information, such as feature descriptions, which could also help structure learning. While recent works explore using Large Language Models (LLMs) to infer causal relations via direct queries, such methods can be unreliable due to a lack of alignment with the actual data. To address these limitations, we propose Causal Ensemble Agent (CEA), a novel framework that aggregates structural insights from statistical discovery experts across different graph levels via linear opinion pooling, and uses an LLM as a meta-referee to dynamically reweight experts when the aggregated confidence is close to the decision boundary, thereby composing an improved and more complete causal graph. Extensive experiments on both synthetic and real-world datasets demonstrate that CEA achieves the strongest overall performance across a wide range of causal discovery methods, highlighting the effectiveness of using LLMs for meta-analysis in causal discovery.

Problem

• causal discovery
• causal structure
• observational data
• domain knowledge
• conflicting results

Innovation

• Causal Discovery
• Large Language Models
• Ensemble Learning
• Hierarchical Causal Structure
• Expert Reweighting