Analyzing the evolution of environmental policy themes and governance stakeholder networks: A computational approach

Environmental policy is a critical tool for countries to establish principles and guidelines for effective environmental governance. This paper explores the evolution of environmental policy themes and stakeholder networks in China’s governance framework over a decade. Based on a dataset of 1,153 policy documents from 31 provinces, this study applies natural language processing for tokenization and utilizes unsupervised and supervised machine learning techniques — Dynamic Topic Modeling for topic detection and Named Entity Recognition for stakeholder extraction. Through spatio-temporal analysis and network analysis, this innovative approach provides new insights into governance dynamics and establishing a novel research paradigm for the academic community. Findings indicate a significant transition from a government-centric approach to a collaborative multi-stakeholder governance model that enhances data-driven decision-making and offers practical recommendations for policymakers. • Introduce DTM and NER as innovative methods for analyzing the evolving landscape of environmental governance. • Examine the valuable co-occurrence networks of policy themes and stakeholders. • Extract governance stakeholders from environmental policy and analyze their value network relationships. • Indicate green technology innovation is integrated into more comprehensive aspects of environmental issues and increasingly shaped by diverse governance actors. • Find a significant transition from a government-centric approach to a collaborative multi-stakeholder governance model.