Towards understanding policy design through text-as-data approaches: The policy design annotations (POLIANNA) dataset

Area of Science:

• Environmental Policy Analysis
• Computational Social Science
• Machine Learning Applications

Background:

• Manual analysis of climate and energy policies is time-consuming and expensive.
• Systematic assessments of policy design are crucial for effective climate change mitigation.
• Existing methods struggle to scale the analysis of extensive policy documentation.

Purpose of the Study:

• To introduce POLIANNA, a novel dataset of annotated European Union (EU) climate and energy policy texts.
• To facilitate the development of supervised machine learning models for automated policy analysis.
• To address the limitations of manual text analysis in policy research.

Main Methods:

• Development of a novel coding scheme based on theoretical policy design concepts.
• Annotation of 20,577 text spans from 18 EU climate change mitigation and renewable energy policies.
• Analysis of inter-annotator agreement to ensure data reliability.

Main Results:

• Creation of the POLIANNA dataset, a valuable resource for policy informatics.
• Demonstration of a method for translating policy design taxonomies into text annotations.
• Establishment of a foundation for developing AI-driven tools for policy analysis.

Conclusions:

• The POLIANNA dataset enables scalable, data-driven analysis of climate and energy policies.
• Automated tools can assist researchers by proposing relevant text segments for manual review.
• This work advances the field of computational policy analysis for climate action.