Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Discovering and Reasoning of Causality in the Hidden World with Large Language Models

Chenxi Liu, Yongqiang Chen, Tongliang Liu, Mingming Gong, James Cheng, Bo Han, Kun Zhang

TL;DR

A new framework termed Causal representatiOn AssistanT (COAT) is developed that incorporates the rich world knowledge of LLMs to propose useful measured variables for CD with respect to high-value target variables on their paired unstructured data and extends the debiased causal inference to unstructured data by discovering an adjustment set.

Abstract

Revealing hidden causal variables alongside the underlying causal mechanisms is essential to the development of science. Despite the progress in the past decades, existing practice in causal discovery (CD) heavily relies on high-quality measured variables, which are usually given by human experts. In fact, the lack of well-defined high-level variables behind unstructured data has been a longstanding roadblock to a broader real-world application of CD. This procedure can naturally benefit from an automated process that can suggest potential hidden variables in the system. Interestingly, Large language models (LLMs) are trained on massive observations of the world and have demonstrated great capability in processing unstructured data. To leverage the power of LLMs, we develop a new framework termed Causal representatiOn AssistanT (COAT) that incorporates the rich world knowledge of LLMs to propose useful measured variables for CD with respect to high-value target variables on their paired unstructured data. Instead of directly inferring causality with LLMs, COAT constructs feedback from intermediate CD results to LLMs to refine the proposed variables. Given the target variable and the paired unstructured data, we first develop COAT-MB that leverages the predictivity of the proposed variables to iteratively uncover the Markov Blanket of the target variable. Built upon COAT-MB, COAT-PAG further extends to uncover a more complete causal graph, i.e., Partial Ancestral Graph, by iterating over the target variables and actively seeking new high-level variables. Moreover, the reliable CD capabilities of COAT also extend the debiased causal inference to unstructured data by discovering an adjustment set. We establish theoretical guarantees for the CD results and verify their efficiency and reliability across realistic benchmarks and real-world case studies.

Discovering and Reasoning of Causality in the Hidden World with Large Language Models

TL;DR

A new framework termed Causal representatiOn AssistanT (COAT) is developed that incorporates the rich world knowledge of LLMs to propose useful measured variables for CD with respect to high-value target variables on their paired unstructured data and extends the debiased causal inference to unstructured data by discovering an adjustment set.

Abstract

Revealing hidden causal variables alongside the underlying causal mechanisms is essential to the development of science. Despite the progress in the past decades, existing practice in causal discovery (CD) heavily relies on high-quality measured variables, which are usually given by human experts. In fact, the lack of well-defined high-level variables behind unstructured data has been a longstanding roadblock to a broader real-world application of CD. This procedure can naturally benefit from an automated process that can suggest potential hidden variables in the system. Interestingly, Large language models (LLMs) are trained on massive observations of the world and have demonstrated great capability in processing unstructured data. To leverage the power of LLMs, we develop a new framework termed Causal representatiOn AssistanT (COAT) that incorporates the rich world knowledge of LLMs to propose useful measured variables for CD with respect to high-value target variables on their paired unstructured data. Instead of directly inferring causality with LLMs, COAT constructs feedback from intermediate CD results to LLMs to refine the proposed variables. Given the target variable and the paired unstructured data, we first develop COAT-MB that leverages the predictivity of the proposed variables to iteratively uncover the Markov Blanket of the target variable. Built upon COAT-MB, COAT-PAG further extends to uncover a more complete causal graph, i.e., Partial Ancestral Graph, by iterating over the target variables and actively seeking new high-level variables. Moreover, the reliable CD capabilities of COAT also extend the debiased causal inference to unstructured data by discovering an adjustment set. We establish theoretical guarantees for the CD results and verify their efficiency and reliability across realistic benchmarks and real-world case studies.
Paper Structure (86 sections, 9 theorems, 21 equations, 44 figures, 11 tables, 3 algorithms)

This paper contains 86 sections, 9 theorems, 21 equations, 44 figures, 11 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Causal Methods for Structured Data
  4. Causal discovery
  5. Causal effect estimation
  6. Causal Methods for Unstructured Data
  7. Causal representation learning
  8. Causal reasoning with LLMs
  9. Causal learning with LLMs
  10. The Causal representatiOn AssistanT Framework
  11. Notation
  12. Active Causal Representation Elicitation
  13. Data
  14. LLM as a representation assistant
  15. Active causal representation elicitation
  16. ...and 71 more sections

Key Result

Proposition 3

Under assumption ass:Faithful_Markov, if condition eq:mutual_info_down:notind holds, then for Markov Blanket ${\mathcal{S}} \subseteq [k+1]$ of $Y$, i.e., $Y \perp\!\!\!\!\perp h_{[k+1] \setminus {\mathcal{S}}}({\bm{X}}) \mid h_{{\mathcal{S}}}({\bm{X}})$, we have the following about conditional mu

Figures (44)

  • Figure 1: An example where PAG is less informative when limited in $\text{MB}(Y)$. (a) is the true causal structure indicating causal directions; (b) and (c) are the partial ancestral graphs. The circle mark ($\circ$) means it is undetermined to be arrow head or tail. The node out of $\text{MB}(Y)$ is colored.
  • Figure 2: Illustration of COAT framework in eliciting the Markov Blanket of the rating score from the paired text reviews. COAT first (a) adopts an LLM to read, comprehend, and relate the rich knowledge about reviews on apples. The LLM needs to propose a series of candidate factors such as apple sizes and smells, along with some meta-information such as annotation guidelines. Based on the candidate factors, COAT then (b) prompts another LLM to annotate the unstructured review into structured data. (c) The CD algorithm then finds causal relations among the factors, and constructs feedback based on samples where the ratings can not be well explained by the existing factors. By looking into the new samples, the LLM is expected to use more related knowledge to uncover the desired causal factors.
  • Figure 3: Illustration of the prompt template for factor proposal in COAT.
  • Figure 4: Illustration of variables that could be discovered with COAT. Let $Y$ be the target variable, and $W$ be a factor that has been discovered, and also assume a latent variable $U=\widehat{{\bm{w}}}({\bm{X}}) \in \text{MB}(Y)$. Conditioning on $W$ facilitates the discovery of $U$.
  • Figure 5: An example where arrow heads cannot be revealed within $\text{MB}^{(N)}(Y)$.
  • ...and 39 more figures

Theorems & Definitions (15)

  • Definition 1: Representation Assistant
  • Proposition 3
  • Definition 4: Ability of LLMs
  • Proposition 5: Characterization for Factor Identification Process
  • Proposition 6: Rate of Convergence
  • Definition 7
  • Proposition 8
  • Corollary 9
  • Definition 10: Generalized adjustment criterion perkovi2018complete
  • Definition 11: Amenability perkovi2018complete
  • ...and 5 more