Populist rhetoric is increasingly prevalent worldwide and affects all countries. Although populism has been identified as a heterogeneous political phenomenon and a multidimensional construct, its detection in the text often remains abstract, limiting the detailed view of populist rhetorical patterns. People-centrism and anti-elitism are often named as one of the core features of populism. To better analyse these dimensions, especially anti-elitism, we extend the idea of Opinion Role Labeling (ORL) to explore negative and positive stances towards the elite and the people. In this talk, I will showcase how natural language processing methods can be utilized to automatically analyse populist rhetoric in texts. Specifically, we will discuss how ORL can be used to identify negative attitudes toward specific groups by detecting group mentions and related stances. We will also delve into the importance of tailoring these models to a certain domain (e.g., parliamentary debates). As a preview of my current research, I will also present how large language models can be applied in this context.
About the series:
The CIVICA Data Science seminar series is an interdisciplinary forum of the CIVICA European University of Social Sciences, an alliance of eight leading higher education and research institutions. The series fosters a community of researchers, practitioners, and policymakers working to address complex societal challenges through the use of novel methodological and data-driven approaches, drawing from machine learning and quantitative methods from biology, sociology, political sciences, economics, and physics, among other disciplines. Shared among the partner institutions of the CIVICA network, this series directly addresses the data science research stream of the CIVICA initiative.
The Zoom link will be sent upon registration.