Abstract:We propose a novel framework that leverages LLMs for full causal graph discovery. While previous LLM-based methods have used a pairwise query approach, this requires a quadratic number of queries which quickly becomes impractical for larger causal graphs. In contrast, the proposed framework uses a breadth-first search (BFS) approach which allows it to use only a linear number of queries. We also show that the proposed method can easily incorporate observational data when available, to improve performance. In addition to being more time and data-efficient, the proposed framework achieves state-of-the-art results on real-world causal graphs of varying sizes. The results demonstrate the effectiveness and efficiency of the proposed method in discovering causal relationships, showcasing its potential for broad applicability in causal graph discovery tasks across different domains.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is **how to efficiently discover causal graphs (Causal Graph)**, especially by using large - language models (LLMs) to achieve this goal. Specifically, the paper proposes a new framework aiming to overcome the efficiency and scalability issues of existing methods when dealing with larger causal graphs. ### Problem Background Traditional causal graph discovery methods usually rely on statistical or continuous optimization techniques. These methods require a large amount of observational data, and as the number of causal variables increases, the computational complexity will grow rapidly. For example, the method based on pairwise queries requires \(O(n^2)\) queries, which becomes impractical when dealing with larger - scale causal graphs. ### The New Method Proposed in the Paper To solve the above problems, this paper proposes a method based on breadth - first search (BFS), combined with large - language models (LLMs) to discover the complete causal graph. The main features of this method include: 1. **Linear query complexity**: By using the BFS strategy, the new method only requires \(O(n)\) queries instead of \(O(n^2)\) queries, thus significantly improving efficiency. 2. **No need for observational data**: Unlike traditional numerical methods, this method does not require additional observational data and only depends on the internal knowledge and metadata of LLMs (such as variable names and descriptions). 3. **Guarantee DAG constraints**: The new method ensures that the generated causal graph is a directed acyclic graph (DAG), which is one of the important characteristics of causal relationships. 4. **Easy to expand**: When observational data is available, this method can easily incorporate this data into consideration to further improve performance. ### Experimental Results The experimental results show that this method has achieved state - of - the - art performance on three real - world causal graphs of different scales, especially on the larger Neuropathic Pain causal graph. Compared with other methods, this method is not only more efficient but also able to handle larger - scale causal graphs. ### Summary In general, this paper proposes a novel and efficient causal graph discovery method. By using the internal knowledge of LLMs and the BFS strategy, it successfully solves the efficiency and scalability problems of existing methods when dealing with larger causal graphs.

Efficient Causal Graph Discovery Using Large Language Models

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

Large Language Models are Effective Priors for Causal Graph Discovery

Large Language Models for Constrained-Based Causal Discovery

CausalGraph2LLM: Evaluating LLMs for Causal Queries

From Query Tools to Causal Architects: Harnessing Large Language Models for Advanced Causal Discovery from Data

Causal Dataset Discovery with Large Language Models

Counterfactual Causal Inference in Natural Language with Large Language Models

Can large language models build causal graphs?

ALCM: Autonomous LLM-Augmented Causal Discovery Framework

Causal Inference Using LLM-Guided Discovery

Multi-Agent Causal Discovery Using Large Language Models

Causal Reasoning and Large Language Models: Opening a New Frontier for Causality

Causal Structure Learning Supervised by Large Language Model

Zero-shot Causal Graph Extrapolation from Text via LLMs

Large Language Models and Causal Inference in Collaboration: A Comprehensive Survey

Causal Inference with Large Language Model: A Survey

Applying Large Language Models for Causal Structure Learning in Non Small Cell Lung Cancer

Causal Reasoning in Large Language Models: A Knowledge Graph Approach

Large Language Model for Causal Decision Making

LLM-initialized Differentiable Causal Discovery