🤖 AI Summary
Traditional causal discovery methods largely ignore semantic context encoded in metadata, while existing LLM-based approaches have not systematically explored causal graph construction in multi-agent settings. Method: We propose MAC, a Multi-Agent Causal discovery framework that pioneers the integration of LLMs into a multi-agent debate paradigm. It features a “Debate Encoding” module that dynamically selects and fuses statistical causal discovery (SCD) algorithms, and a “Meta-Debate” module that refines causal graphs using metadata. A novel Meta Fusion mechanism enables reversible mapping and joint optimization between causal graphs and metadata. Contribution/Results: MAC is the first framework to unify statistical inference and semantic knowledge within a multi-agent architecture. Evaluated on five benchmark datasets, it significantly outperforms state-of-the-art SCD methods and LLM-based baselines, achieving new SOTA performance with improved accuracy and robustness of learned causal graphs.
📝 Abstract
Causal discovery aims to identify causal relationships between variables and is a critical research area in machine learning. Traditional methods focus on statistical or machine learning algorithms to uncover causal links from structured data, often overlooking the valuable contextual information provided by metadata. Large language models (LLMs) have shown promise in creating unified causal discovery frameworks by incorporating both structured data and metadata. However, their potential in multi-agent settings remains largely unexplored. To address this gap, we introduce the Multi-Agent Causal Discovery Framework (MAC), which consists of two key modules: the Debate-Coding Module (DCM) and the Meta-Debate Module (MDM). The DCM begins with a multi-agent debating and coding process, where agents use both structured data and metadata to collaboratively select the most suitable statistical causal discovery (SCD) method. The selected SCD is then applied to the structured data to generate an initial causal graph. This causal graph is transformed into causal metadata through the Meta Fusion mechanism. With all the metadata, MDM then refines the causal structure by leveraging a multi-agent debating framework. Extensive experiments across five datasets demonstrate that MAC outperforms both traditional statistical causal discovery methods and existing LLM-based approaches, achieving state-of-the-art performance.