Geospatial Event Detection in the Twitter Stream


The rise of Social Media in the last years has created huge streams of information that can be very valuable in a variety of scenarios. However, what precisely these scenarios are and how the data streams can efficiently be analyzed for each scenario is still largely unclear.

In this talk, we present a system for geo-spatial event detection on Social Media streams. It monitors all posts on Twitter issued in a given geographic region, store them in MongoDB and usees a spatial index to identify places that show a high amount of activity. In a second processing step, we analyze the resulting spatio-temporal clusters of posts with a Machine Learning component in order to detect whether they constitute real-world events or not. In our experiments, we focus on the New York metropolitan area and we detect events as diverse as house fires, on-going baseball games, parties, traffic jams, Broadway premiers, conferences, tech gatherings and demonstrations.

We report on the algorithms and technology we use to efficiently analyze the Twitter data stream in near real-time and address uses cases of our system.

About the speaker: 
Michael Kaisser is currently leading AGT International's Social Media Analytics R&D team in Berlin, Germany. He is also the co-founder of txtData, which provides consulting services on Natural Language Processing, Machine Learning, Information Retrieval and related topics . In the past he worked at Microsoft Bing's Search Technology Centre Munich on search core relevance as a Program Manager for Query Alterations and Spell Correction. Michael holds a PhD from the University of Edinburgh, where he worked on Question Answering and developed methods to push web-search beyond the usual ten blue links. He likes parsnips but finds tennis rather dull.

Schedule info

Time slot: 
4 June 12:20 - 13:05