Improved Burst Based Real-Time Event Detection Using Location Dependent Corpora

This paper describes an approach to detect automatically events in real-time by analyzing big data streams coming from a social network. The social network Twitter will be used for experiments because of its easy to use API and messages having positioning data. The described event detection algorithm is a burst-based event detection methods extended by a dynamically adapted reference corpus. The usage of location dependent corpora allows also detecting the location of the found events.