DIGSSCORE Seminar: Quantitative text analysis: from bag-of-words to word embeddings

Endre Tvinnereim, Professor at the Department of Government at University of Bergen, will present for us today. His presentation is titled "Quantitative text analysis: from bag-of-words to word embeddings".

The event is in a hybrid format, you are welcome to join us for lunch from the Corner room at DIGSSCORE. Food is provided on a first-come first-served basis. Zoom link for digital attendance.

Abstract:

Quantitative text analysis (QTA) is becoming increasingly popular, as it allows researchers to analyze large bodies of text. For survey researchers, QTA is particularly interesting for the analysis of textual responses to open-ended survey questions. In this talk, I will discuss developments in QTA, beginning with bag-of-words methods such as keyness analysis, word2num, wordfish, and structural topic modeling. I will then introduce BERTopic, an approach using insights from language models trained on large amounts of textual data, and provide an example based on a multilingual data set drawn from the European Perceptions of Climate Change data set. I will also discuss matters relating to validation of model results and QTA applied to small languages.