Place: 4274 Chamberlin (refreshments will be served)
Speaker: Adel Ardalan, UW Department of Computer Science
Abstract: Twitter has gone viral over the past few years, both in terms of the volume of posted messages and the number of users. People talk about real-world events on Twitter before events start, during the events and after they are finished. In many applications, it is desirable to monitor the Twitter stream and detect the interesting events people are discussing. In this talk, we define and motivate the event detection problem and suggest a model for detecting emerging, dynamic, interesting events. We discuss the challenges we have faced in building a system to generate candidate events based on the aforementioned model in a timely manner and how we have addressed them. Our experiments show promising results both in terms of accuracy of our detection and also the response time of the system.<br>