This is the public repository of our project Tracking the Infodemic, concerned with identifying conspiracy narratives in German text and tracing their distribution on social media, analysing typical linguistic patterns and discursive strategies, and much more.
In addition to our collected talks and presentations, you can find a list of 63 fine-grained categories (including examples from Telegram) for annotation purposes here. A manually annotated sample is also available, split into training, development and test data.
The folder schwurbelocity contains data originally annotated by three independent annotators on different aspects/dimensions indicating "Geschwurbel", i.e. problematic texts often containing one or more of the above-mentioned narratives. Strong disagreements between annotators were manually adjudicated, otherwise, we use the mean as an automatic adjudication. Annotations guidelines are also available. More information about this data can be found in our short paper "Dimensions of Drivel in German Telegram Posts Manual Annotation and Predictive Power" in the Proceedings of the 12th International Conference on CMC and Social Media Corpora for the Humanities.