Using Social Media to Estimate the Audience Sizes of Public Events for Crisis Management and Emergency Care

Public events such as soccer games, concerts, or street festivals attract large crowds of visitors. In an emergency situation, estimations about current events and their numbers of visitors are important to be able to react early and effectively by performing adequate countermeasures. Previous research has proposed ap-proaches to detect events like accidents and catastrophes by relying on user-generated content and reporting event-related information. To be proactive in case of an emergency, it is important to know what is happening in direct proximity, even if it is not yet affected by the catastrophe. Therefore, information about on-going events and numbers of visitors in the surrounding environment is indis-pensable. We develop a system design that allows collecting and merging event-related information from social media to provide estimations of the audience siz-es. We illustrate the potential of our approach by estimating the number of visi-tors of soccer games, fairs, street festivals, music festivals, and concerts, and by comparing it to the real numbers of visitors. Our results indicate that matching event-related user-generated content leads to improvements of the estimations. Finally, we demonstrate the usefulness of the system in a recent crisis scenario