ParlaCAP: Dataset for tracking political agenda-setting across European parliaments

The ParlaCAP dataset consists of 8 million speeches from 28 European national and regional parliaments, with each speech coded with the sentiment expressed (ParlaSent coding from negative, over neutral, to positive) and the topic discussed (Comparative Agendas Project coding with 22 topics), and rich metadata on the speakers, parties and democracies. The dataset is an extension of the ParlaMint 5.0 dataset, which was primarily focused on the transcripts of parliamentary speeches and their metadata. The ParlaCAP dataset extends the ParlaMint dataset via the “text as data” paradigm by automatically coding topics and sentiment for each speech, simplifying the data to a tabular form, and thereby empowering social science research on agenda setting and negativity in political discourse across a broad set of parliaments. For automatic coding, multilingual transformer models were used, with the ParlaCAP model for topic, and the ParlaSent model for sentiment.