Högre seminarium: Måns Magnusson

During the last years, large textual corpora for research, such as parliamentary proceedings, newspaper corpora, or court cases, are becoming increasingly available in digital formats. 

These corpora and new computational methods and statistical approaches enable researchers to study these large materials in new ways and pose new research questions.

In this talk, I will present three popular statistical approaches for textual data, (1) topic models, (2) probabilistic word embeddings and (3) transformer neural networks for supervised machine learning, and how they can be used in applied research for large textual materials.