Computational Advances in the Narrative Policy Framework: Review and Outlook

There is growing interest in applying computational methods to studying policy narratives through the lens of the Narrative Policy Framework (NPF). Given that NPF is rooted in qualitative content analysis, turning it into an automated task has raised several challenges. The goal of this literature review is to provide an overview of the current state of research on the computational advances in the NPF and to create an understanding of the requirements for big data applications that account for recent developments in language technology. NPF scholars distinguish between narrative form and content, the former referring to narrative elements (such as setting, characters or plot) and the latter referring to the belief systems and strategies of the narrative. To date, different computational methods used in NPF applications mainly targeted narrative elements, relying on such Natural Language Processing (NLP) techniques as named entity recognition, sentiment analysis and network analysis. We further critically examine the strengths and limitations of these techniques and discuss their potential for NPF research. Finally, we build on the recent advancements in digital humanities to explore the potential of Large Language Models (LLMs) to contribute to the analysis of narrative content. The study also briefly explores the challenges of using fully automated methods in the NPF, such as potential algorithmic biases. We highlight the potential for future research in this area and the need for interdisciplinary collaborations between computer scientists, social scientists, and policy analysts to advance NPF research and its Innovative applications.