A Comparison of Language Processing Models in Political Analysis : Evidence from Sweden
This study evaluates two different natural language processing tech-niques: the normalised co-occurrence (PMI) versus neural networks. Incontrast to most previous studies, the focus is on the context of a propor-tional representation system – Sweden –, where the parties in parliamenttend to form coalitions. We test the models by collecting data from thenational parliament (Swedish Riksdag) and usi
