Transforming literature into causal system maps: A policy analysis pipeline for decarbonizing the power sector

The use of causal systems mapping in interdisciplinary and policy research has increased in recent years. Causal system maps typically rely on stakeholder opinion for their creation. While this works well, it does not fully leverage the available data and can be time-consuming. For most topics, there is an abundance of text data in easily identifiable academic literature, grey literature, and policy documents. Using this data to support causal systems mapping exercises has the potential to make them more comprehensive and connected to evidence. In this paper, we develop a Natural Language Processing (NLP)-based pipeline that uses literature text to construct causal system maps. Using power sector decarbonisation policies as an example, and comparing the results with a related participatory exercise, we explore suitable techniques, strategies that might speed up mapping exercises, and potential risks. The resulting NLP-generated map captures familiar factors and logical individual relationships, all of which are traceable to original references, ensuring transparency and verifiability. However, its overall structure tends to reflect patterns of attention in the literature rather than underlying causal mechanisms, and it overemphasises connections directly between policies and outcomes, rather than longer, more realistic causal chains. By contrast, the participatory map has a clearer and more purpose-driven structure.