# Zero-shot Causal Graph Extrapolation from Text via LLMs

arXiv: https://arxiv.org/abs/2312.14670 · HTML: https://mlsanity.com/paper/2312.14670/

## Key facts

- Cited by 15 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 15 references that are in the map.
- First posted on arXiv 22 Dec 2023.

## Builds on (most cited first, 9 of 9)

- [A Survey on Knowledge Graphs: Representation, Acquisition and Applications](https://mlsanity.com/paper/2002.00388/) (2020)
- [Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm](https://mlsanity.com/paper/2102.07350/) (2021)
- [Deep Learning for Medical Image Analysis](https://mlsanity.com/paper/1708.08987/) (2017)
- [Can Large Language Models Infer Causation from Correlation?](https://mlsanity.com/paper/2306.05836/) (2023)
- [Comparative Study of CNN and RNN for Natural Language Processing](https://mlsanity.com/paper/1702.01923/) (2017)
- [A Survey on Text Classification: From Shallow to Deep Learning](https://mlsanity.com/paper/2008.00364/) (2020)
- [Large Language Models and Knowledge Graphs: Opportunities and Challenges](https://mlsanity.com/paper/2308.06374/) (2023)
- [Understanding Causality with Large Language Models: Feasibility and Opportunities](https://mlsanity.com/paper/2304.05524/) (2023)
- [Approximating Counterfactual Bounds while Fusing Observational, Biased and Randomised Data Sources](https://mlsanity.com/paper/2307.16577/) (2023)

## Built on by (most cited first, 15 of 15)

- [ALCM: Autonomous LLM-Augmented Causal Discovery Framework](https://mlsanity.com/paper/2405.01744/) (2024)
- [Large Language Models and Causal Inference in Collaboration: A Survey](https://mlsanity.com/paper/2403.09606/) (2024)
- [Causality for Large Language Models](https://mlsanity.com/paper/2410.15319/) (2024)
- [End-To-End Causal Effect Estimation from Unstructured Natural Language Data](https://mlsanity.com/paper/2407.07018/) (2024)
- [Causally Reliable Concept Bottleneck Models](https://mlsanity.com/paper/2503.04363/) (2025)
- [Structure-Augmented Reasoning Generation](https://mlsanity.com/paper/2506.08364/) (2025)
- [Score-informed Neural Operator for Enhancing Ordering-based Causal Discovery](https://mlsanity.com/paper/2508.12650/) (2025)
- [StructSynth: Dependency Graphs as Generation Plans for Low-Data Tabular Synthesis with Language Models](https://mlsanity.com/paper/2508.02601/) (2025)
- [Designing Computational Tools for Exploring Causal Relationships in Qualitative Data](https://mlsanity.com/paper/2602.06506/) (2026)
- [iTAG: Inverse Design for Natural Text Generation with Accurate Causal Graph Annotations](https://mlsanity.com/paper/2604.06902/) (2026)
- [InsightFlow: LLM-Driven Synthesis of Patient Narratives for Mental Health into Causal Models](https://mlsanity.com/paper/2604.12721/) (2026)
- [ORCA: An End-to-End Interactive Copilot for Optimized Root Cause Analysis](https://mlsanity.com/paper/2605.27022/) (2026)
- [Implicit Causal Graph Construction in Text via Chain Discovery](https://mlsanity.com/paper/2606.07525/) (2026)
- [CausalSteward: An Agentic Divide-Conquer-Combine Copilot for Causal Discovery](https://mlsanity.com/paper/2607.01936/) (2026)
- [DKCD: Domain Knowledge-Enhanced Causal Discovery from Unstructured Data](https://mlsanity.com/paper/2607.09348/) (2026)

## Abstract (arXiv)

We evaluate the ability of large language models (LLMs) to infer causal relations from natural language. Compared to traditional natural language processing and deep learning techniques, LLMs show competitive performance in a benchmark of pairwise relations without needing (explicit) training samples. This motivates us to extend our approach to extrapolating causal graphs through iterated pairwise queries. We perform a preliminary analysis on a benchmark of biomedical abstracts with ground-truth causal graphs validated by experts. The results are promising and support the adoption of LLMs for such a crucial step in causal inference, especially in medical domains, where the amount of scientific text to analyse might be huge, and the causal statements are often implicit.