Estonian Parliament Speeches as Source for Content Analysis
ParlaMint is a CLARIN-supported project to create uniformly annotated multilingual corpora of parliamentary sessions. The Estonian team joined the project for Stage 2, where the XML schema and most of the tools were worked out already.
The paper will outline some of the tools that are at our disposal for the analysis of the style and content of such corpora. We will introduce first versions of sentiment and populism classification models that have been specifically trained on political texts and explore how topic models can be used to trace the evolution of the party system through the analysis of parliamentary speech.