Actions for Event Detection in Twitter Data : A Hidden Markov Model-Based Change Point Algorithm

Email RIS file
Bookmark
Report an Issue

Event Detection in Twitter Data : A Hidden Markov Model-Based Change Point Algorithm

Author
Osotsi, Ame
Published
[University Park, Pennsylvania] : Pennsylvania State University, 2016.
Physical Description
1 electronic document
Additional Creators
Li, Qunhua
Access Online

Availability

I Want It
I Want It

Finding items...

Graduate Program
Restrictions on Access
Open Access.
Summary
Twitter is a popular microblogging platform that displays real-time status updates from over 140 million users a day. The users post about anything, from daily life events to important global events. We attempt to analyze this rich source of user-generated data using hidden Markov models, which have been very successful in describing time series observations. The aim of this project is to quantify how conversation in Twitter evolve in response to two major events: an unexpected school shooting, and the Super Bowl. We use a hidden Markov model-based change point algorithm. This thesis first introduces the data and the hidden Markov models underlying the change point algorithm. We then describe the change point algorithm and related problems such as finding confidence intervals, model selection, and computing summaries. Finally, we show results on the two datasets and propose future avenues of research.
Other Subject(s)
Genre(s)
Dissertation Note
M.S. Pennsylvania State University 2016.
Reproduction Note
Library holds archival microfiches negative and service copy. 1 fiche. (Micrographics International, 2016)
Technical Details
The full text of the dissertation is available as an Adobe Acrobat .pdf file ; Adobe Acrobat Reader required to view the file.
View MARC record | catkey: 18372798