Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
MS_Thesis_Submission_V3.pdf (501.42 KB)
ETD Abstract Container
Abstract Header
Forecasting event outcomes from user predictions on Twitter
Author Info
Swamy, Sandesh
ORCID® Identifier
http://orcid.org/0000-0001-6069-580X
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1492692142585459
Abstract Details
Year and Degree
2017, Master of Science, Ohio State University, Computer Science and Engineering.
Abstract
Social media websites like Twitter provide an ideal environment for people to express their opinions and beliefs about upcoming events. When an event like the Oscars is around the corner, users on social media websites try to make predictions about potential winners at the event. For example, Leo will finally win the Oscar for Revenant! suggests that the author of the tweet believes that Leonardo DiCaprio will win at the Oscars. Can user predictions on Twitter be an indicator of the actual outcome of these events? This work focuses on exploring the answer to this question by using the concept of veridicality. In social media, we find tweets of varying degrees of veridicality. Before the 2017 Oscars, we see tweets about Moonlight such as Moonlight will win Best Picture! vs. Moonlight will lose to La La Land. and I need Moonlight to win!. The first tweet indicates that the author believes, with certainty, that Moonlight will win. This is positive veridicality towards an outcome. The second tweet indicates the opposite - Moonlight not winning - and is an expression of negative veridicality towards the event. We build a classifier which can automatically detect the veridicality of a given tweet with high precision. Further, by using this classifier to detect tweets with positive veridicality, it can be shown that we can make predictions on outcomes of speculative events much better than sentiment and volume-based predictions made on the same events. Our analysis also allows us to retrospectively predict surprise outcomes in such speculative events. We also show that our method can be used to assess the reliability of accounts which are making predictions on such events.
Committee
Alan Ritter (Advisor)
Marie-Catherine de Marneffe (Advisor)
Wei Xu (Committee Member)
Pages
29 p.
Subject Headings
Computer Engineering
;
Computer Science
;
Language
;
Linguistics
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Swamy, S. (2017).
Forecasting event outcomes from user predictions on Twitter
[Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1492692142585459
APA Style (7th edition)
Swamy, Sandesh.
Forecasting event outcomes from user predictions on Twitter.
2017. Ohio State University, Master's thesis.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1492692142585459.
MLA Style (8th edition)
Swamy, Sandesh. "Forecasting event outcomes from user predictions on Twitter." Master's thesis, Ohio State University, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=osu1492692142585459
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1492692142585459
Download Count:
1,248
Copyright Info
© 2017, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.