Network Analysis of German COVID-19 Related Discussions on Telegram

ABSTRACT

We present an effective way to create a dataset from relevant channels and groups of the messenger service Telegram, to detect clusters in this network, and to find influential actors. Our focus lies on the network of German COVID-19 sceptics that formed on Telegram along with growing restrictions meant to prevent the spreading of COVID-19. We create the dataset by using a scraper based on exponential discriminative snowball sampling, combining two different approaches. We show the best way to define a starting point for the sampling and to detect relevant neighbouring channels for the given data. Community clusters in the network are detected by using the Louvain method. Furthermore, we show influential channels and actors by defining a PageRank based ranking scheme. A heatmap illustrates the correlation between the number of channel members and the ranking. We also examine the growth of the network in relation to the governmental COVID-19 measures. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.