Event Detection in Social Media

Event detection deals with finding breaking news in Social Media (typically Twitter). Examples of breaking news include the Twitter account of AP news being hacked (resulting in the Dow Jones plunging 145 points), the killing of Osama bin Laden, forest fires, traffic jams, celebraties dying, people inciting others to riot and more. Breaking news ranges from the most dramatic to the most mundane.

The main goal is to find all breaking news, as quickly as possible. With upwards of 143k Tweets per second (500 million per day), most of which is irrelevant, the challenges include developing machine learning methods that scale well and are capable of finding breaking news in a sea of garbage.

The project is a collaboration between Miles Osborne, Victor Lavrenko from the School of Informatics, University of Edinburgh, and Iadh Ounis, and Craig MacDonald from the School of Computer Science, University of Glasgow.

Papers

Scalable Distributed Event Detection for Twitter pdf
Richard McCreadie, Craig Macdonald, Iadh Ounis, Sasa Petrovic, Miles Osborne.
2013 IEEE International Conference on Big Data. Santa Clara US. October 2013.

Can Twitter replace Newswire for breaking news? pdf
Sasa Petrovic, Miles Osborne, Richard McCreadie, Craig Macdonald, Iadh Ounis, Luke Shrimpton.
ICWSM, Boston US. July 2013.

Bieber no more: First Story Detection using Twitter and Wikipedia pdf
Miles Osborne, Sasa Petrovic, Richard McCreadie, Craig Macdonald, Iadh Ounis.
ICWSM, Boston US. July 2013.

Streaming First Story Detection with application to Twitter pdf
Sasa Petrovic, Miles Osborne and Victor Lavrenko.
NAACL, Los Angeles, USA. June 2010.

Media Coverage

Resources and Links