Speaker: Xinyue Wang
Title: Capturing Content and Topics of Large-scale Events by Mining Twitter Streams
Abstract:
The widespread use of Microblogging services, such as Twitter, makes them a valuable tool to correlate people’s personal opinions about popular public events. Current Twitter based event monitoring tends to investigate events based on the content retrieved by a set of pre-defined keywords. Consequently, the accuracy of results depends on the microblog feeds being analysed. Whereas, keywords are hard to specify for the event microblog feeds retrieval. In this talk, I’ll introduce a novel adaptive crawling approach that tries to break this dead-lock. The proposed approach continuously monitors the Twitter stream and measures the similarity between any emerging potential keywords and the pre-defined keywords. Top rated ones are identified as event topics and added as new keywords in real time. The evaluation is conducted by comparing the proposed approach with a (non-adaptive) baseline one in a real-world event. Results show that the adaptive crawling approach not only retrieves extra amount of event content, but also accurately identifies keyword topics that may emerge in the midst of events, as they unfold.
Date: 14th May, 2014.
Time: 14.00-15.00 hrs
Venue: QMUL Mat 1.03