Predicting the number of COVID-19 cases using data from Twitter and Google

In a joint project with Frankfurt University, UKP co-authors show how the prediction of new COVID-19 cases can be improved using Twitter and Google data. Because impactful decisions of politics rely on the estimates of future corona cases, accurate predictions are crucial In their study, the team shows how predictions from John Hopkins University of COVID-19 cases in Germany differ from the official numbers by 79%. By including relevant Google search queries and tweets, the number of new corona cases can potentially be predicted better than by using data from JHU alone. Find the full study here.