Party Manifesto

Party manifesto analysis uses computational methods, primarily natural language processing (NLP) techniques like transformer models (e.g., BERT, DistilBERT), to automatically quantify political positions and sentiments expressed in party manifestos across multiple languages. Current research focuses on improving the accuracy and robustness of these models, particularly addressing challenges related to cross-lingual transfer and the handling of long texts, while also exploring the relationship between expressed emotions, party status (incumbent vs. opposition), and ideological similarity. This research contributes to a more nuanced understanding of political discourse and provides valuable tools for political scientists to analyze large-scale datasets efficiently, potentially informing comparative political studies and improving the accuracy of political forecasting.