Causal Discovery with Language Models as Imperfect Experts

TL;DR

This paper investigates how to enhance causal discovery by integrating potentially imperfect expert knowledge, including language models, to improve causal graph identification beyond traditional methods.

Contribution

It introduces strategies for correcting expert-provided causal orientations using consistency checks and demonstrates this approach with a large language model as an imperfect expert.

Findings

Language models can serve as imperfect experts for causal discovery.

Consistency-based methods improve causal graph accuracy.

Real data case study validates the approach.

Abstract

Understanding the causal relationships that underlie a system is a fundamental prerequisite to accurate decision-making. In this work, we explore how expert knowledge can be used to improve the data-driven identification of causal graphs, beyond Markov equivalence classes. In doing so, we consider a setting where we can query an expert about the orientation of causal relationships between variables, but where the expert may provide erroneous information. We propose strategies for amending such expert knowledge based on consistency properties, e.g., acyclicity and conditional independencies in the equivalence class. We then report a case study, on real data, where a large language model is used as an imperfect expert.