Zero-shot Causal Graph Extrapolation from Text via LLMs
This addresses the challenge of extracting causal graphs from vast scientific text, particularly in medical domains where causal statements are often implicit.
The paper tackles the problem of inferring causal relations from natural language using large language models (LLMs), showing competitive performance on pairwise relations without explicit training. It extends this to extrapolating causal graphs via iterated pairwise queries, with promising results on a biomedical abstract benchmark validated by experts.
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.