The aim of the research project is to measure consensus and polarization in the positions about the measures to combat the COVID-19 pandemic of different social groups (science, politics, media, population) on Twitter. With the help of innovative methods from the field of Natural Language Processing (NLP), opinions will be automatically identified and opinion dynamics will be statistically modeled with the help of time-series analytical methods to explore causes and developments of a polarization in society.

Specifically, the following questions will be answered: How did different social groups (e.g., politicians or the media) and subgroups (e.g., different parties and media with different editorial lines) evaluate the Corona measures, how did this change over time, and how did the positions of the different groups mutually influence each other? Moreover, since the NLP models developed here can be applied to future crises, the project enables to identify general patterns in the emergence of consensus and polarization in crises and the establishment of a societal early warning system for divisive tendencies. In addition to this, the project also pursues two methodological objectives: First, comparisons of the expressions of opinion measured on Twitter with representative population surveys are intended to provide information on how well the discourse on Twitter is suited as an indicator of public opinion. Second, the innovative NLP methods can be applied to other question in social science.

KoPoCoV is a joint undertaking of TU Darmstadt together with Johannes Gutenberg-Universität Mainz (Prof. Dr. Marcus Maurer, Dr. Simon Kruschinski)