A sentiment-based approach to measuring multidimensional party positions using transformer
Political Science Research and Methods · 19.11.2026 · 1-17
DOI: 10.1017/psrm.2026.10145 ↗Kurz gefasst
Fachartikel in der Zeitschrift Political Science Research and Methods, online erschienen am 19.11.2026. Verfasst von Hung H. V. Nguyen. Die Arbeit nennt 70 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Deutsche Forschungsgemeinschaft.
Abstract (Originalsprache)
Existing automated position scaling models from texts rely on the conflict-driven assumption, forcing models to search the linguistic space for differentiating features between political subjects. This leads models to conflate rhetorical styles with political sentiments when subjects are not inherently conflictual. This article introduces ContextScale, a sentiment-based position scaling framework leveraging pre-trained Transformer models to predict policy topics and political sentiments and measure topic-specific position scores across multiple policy dimensions at the sequence level. ContextScale offers an alternative way to gauge party positions, focusing on how political texts express attitudes toward policy issues rather than on contrast between actors. Validity tests demonstrate its ability to capture nuances in party positioning and its adaptability across languages, text domains, and theoretical models.
Förderung
- Deutsche Forschungsgemeinschaft
Zitieren
APA
Nguyen, H. H. V. (2026). A sentiment-based approach to measuring multidimensional party positions using transformer. Political Science Research and Methods, 1-17. https://doi.org/10.1017/psrm.2026.10145
Vancouver
Nguyen HHV. A sentiment-based approach to measuring multidimensional party positions using transformer. Political Science Research and Methods. 2026:1-17. doi:10.1017/psrm.2026.10145
BibTeX
@article{Nguyen2026,
author = {Nguyen, Hung H. V.},
title = {A sentiment-based approach to measuring multidimensional party positions using transformer},
journal = {Political Science Research and Methods},
year = {2026},
pages = {1--17},
doi = {10.1017/psrm.2026.10145}
}
EUVIMED