Optimizing Topic Modelling For Comments On Social Networks: Reactions To Science Communication On Covid

Channels on social media which disseminate scientific information ot the public have been of great importance during the COVID crisis. The authors of videos and texts as well as researchers who study the impact of scientific information online are interested in the reactions on the design of such information resources. This study aims to obtain insights into the perception of scientific information for the public. This is related to the relationship between the information behaviour of individuals and science communication (e.g. through videos or texts) during the COVID-19 crisis. To analyze the reactions to scientific information for the public, we selected Twitter users who are doctors, researchers, science communicators or represent research institutes, processing their replies for 20 months from the beginning of the pandemic, and performing topic modeling on the textual data. The goal was to find topics that relate to the theme of reactions to scientific communication. We present a method that filters such reactions of consumers of science communication. Various topic modelling experiments show that topics can support the search for relevant online reactions, defined by sentences such as: "This was very informative! Thanks".