About this Event

I will talk about how in the social sciences (primarily political and communication science), neural NLP is being used to measure constructs of interest that have previously been measured by hand coding. I illustrate this with results from experiments on parliament speeches and party manifestos from the last legislative period in Germany (2017-2021). One key issue I want to discuss (and where social science and computer science sometimes have a different understanding) is measurement validity.

Patrick Schwabl is a Ph.D. student at LMU Munich. He works at the chair for Computational Communication Research of Professor Mario Haim. His research uses word embeddings to measure social science constructs like political positions or societal cleavages in textual data.