Replication Package for Preprint: "Using vision-language models to extract network data from images of system maps" (White and Barbrook-Johnson, 2025)

This repository provides the code and data required to replicate our analysis of how vision language models (VLMs) perform when tasked to extract causal information from images of systems maps. Original preprint: White, J. and Barbrook-Johnson, P. (2025). Using vision-language models to extract network data from images of system maps. Institute for New Economic Thinking Working Paper Series, University of Oxford.本代码仓库提供了复现本次分析所需的代码与数据集，本次分析旨在探究视觉语言模型（Vision Language Models，VLMs）在被要求从系统图图像中提取因果信息时的表现。 原始预印本： White, J. 与 Barbrook-Johnson, P.（2025）：《利用视觉语言模型从系统图图像中提取网络数据》，牛津大学新经济思想研究所工作论文系列。