The German-Speaking Twitter Community Reference Data Set

News providers and politicians increasingly publish and disseminate their content on online social media to reach broader audiences effectively. Directed by ubiquitous mobile use, the majority of individuals reportedly consume daily news directly on these platforms, mainly in an incidental manner. This bears many risks of misconceptions and misinformation: Social media users tend to extend unwarranted trust in posts that are distributed by contacts on the platform and therefore have difficulties evaluating the credibility and trustworthiness of information and its sources. Reduced political proficiency and social understanding have been reported as directed results, as well as the risk of succumbing to partisan echo chambers, user groups amplify and reinforce their own beliefs due to almost exclusive exposition. Measuring and understanding these phenomena requires analysis of the user behavior on these platforms, and a virtually complete data set of one representative community. We focus on Twitter and present collection techniques to obtain a complete data set of specified sub-groups of its users, with the example of the German-tweeting community, in this paper. We show how to collect a representative snapshot of all tweets pertaining to this community over the period of two months. The resulting sample includes 77 million tweets and 6.9 million users. We validate the sample with exhaustive evaluations, and identify the notable impact of political events, such as the 2019 European Parliament election.