Causal Discovery with Language Models as Imperfect Experts

Stephanie Long, Alexandre Piché, Valentina Zantedeschi, Tibor Schuster, Alexandre Drouin
7/5/2023
cs.AIcs.CLcs.LG

Official code: StephLong614/Causal-disco-LLM-imperfect-experts (17 stars, last commit 2024-06-07).

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.

View on arXivView PDF

Code Implementations(1)

StephLong614/Causal-disco-LLM-imperfect-expertsOfficial100%
172Shell, PythonApr 2, 20232 years ago

Similar Papers

Leveraging Large Language Models for Causal Discovery: a Constraint-based, Argumentation-driven Approach

89% match2026

Learning to Defer for Causal Discovery with Imperfect Experts

88% match2025

Large Language Models for Constrained-Based Causal Discovery

88% match1 repo132024

Causal Graph Discovery with Retrieval-Augmented Generation based Large Language Models

85% match2024

Can We Utilize Pre-trained Language Models within Causal Discovery Algorithms?

85% match2023

ALCM: Autonomous LLM-Augmented Causal Discovery Framework

85% match2024

Cite this paper

@article{long2023causal,
  title  = {Causal Discovery with Language Models as Imperfect Experts},
  author = {Stephanie Long and Alexandre Piché and Valentina Zantedeschi and Tibor Schuster and Alexandre Drouin},
  year   = {2023},
  eprint = {2307.02390},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2307.02390v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2307.02390.svg)](https://codeofpaper.com/paper/2307.02390)

Discussion