Twitter in Academic Conferences:Usage, Networking and Participation over Time ACM Conference on Hypertext and Social Media 2014 ---- http://dl.acm.org/citation.cfm?doid=2631775.2631826 http://dx.doi.org/10.1145/2631775.2631826 ---- Xidao Wen, University of Pittsburgh Yu-Ru Lin, University of Pittsburgh Christoph Trattner, Know-Center Denis Parra, Pontificia Universidad Católica de Chile

1. Twitter in Academic Conferences:
Usage, Networking and Participation
over Time
Xidao Wen, University of Pittsburgh
Yu-Ru Lin, University of Pittsburgh
Christoph Trattner, Know-Center
*Denis Parra, Pontificia Universidad Católica de Chile
ACM International Conferencia on Hypertext and Social Media,
Santiago, 2014
*Presenter

2. Outline
• Context & Motivation of this research
• Research Questions
• Question 1: Communication Types Over Time
• Question 2: Network Evolution over Time
• Question 3: Predicting Re-visiting over Time
• Conclusions and Future Work

3. Context
• An approach to Social Computing Research that I
heard in a talk by Bob Kraut to study online social
communities*
*(More Precisely, what I recall from that talk)
Robert Kraut
Carnegie Mellon University

4. Context
• An approach to Social Computing Research that I
heard in a talk by Bob Kraut to study online social
communities
Understand
Communities
(behavior)
1st Step

5. Context
• An approach to Social Computing Research that I
heard in a talk by Bob Kraut to study online social
communities
Understand
Communities
(behavior)
1st Step
Modify/Enhance
Communities
Behavior
2nd Step

6. In this work, we study Twitter at Conferences
• … and study what conference attendees are currently using, and how that
usage has evolved over time: Twitter
Understand
Communities
(behavior)
• How groups of people interact with each other on Twitter during academic
conferences, In Proceedings of the 2014 ACM Conference on Computer
Supported Cooperative Work (CSCW 2014), ACM, Baltimore, Maryland,
USA

7. Many studies doing similar small-sized
analysis
• C. Ross, M. Terras, C. Warwick, and A. Welsh, “Enabled backchannel:
conference twitter use by digital humanists,” Journal of Documentation,
vol. 67, no. 2, pp. 214–237, 2011.
• W. Reinhardt, M. Ebner, G. Beham, and C. Costa, “How people are using
twitter during conferences,” Creativity and innovation Competencies on
the Web, p. 145, 2009.
• M. Ebner, “Introducing live microblogging: How single presentations can be
enhanced by the mass.,” Journal of research in innovative teaching, vol. 2,
no. 1, 2009.
• J. Letierce, A. Passant, J. G. Breslin, and S. Decker, “Using twitter during an
academic conference: The# iswc2009 use-case.,” in ICWSM, 2010.

8. Many studies doing similar small-sized
analysis
• C. Ross, M. Terras, C. Warwick, conference twitter use 1. by All of and these A. Welsh, works
“Enabled backchannel:
digital humanists,” Journal of Documentation,
vol. 67, no. 2, pp. 214–consider 237, 2011.
only a few
• W. Reinhardt, M. Ebner, G. Beham, and C. Costa, “How people are using
twitter during conferences,” Creativity and innovation Competencies on
the Web, p. 145, conferences 2009.
in specific
• M. Ebner, “Introducing live microblogging: domains.
How single presentations can be
enhanced by the mass.,” Journal of research in innovative teaching, vol. 2,
no. 1, 2009.
• J. Letierce, A. Passant, J. G. Breslin, and S. Decker, “Using twitter during an
academic conference: The# iswc2009 use-case.,” in ICWSM, 2010.
2. None of them have
studied them over time

9. This Paper: Twitter in Conferences over Time
• 16 conferences over 5 years in CS & IS

10. This Paper: Twitter in Conferences over Time

11. This Paper: Twitter in Conferences over Time
• RQ1: Do users use Twitter more for socializing with peers or
for information sharing during conferences? How has such
use of Twitter during conferences changed over the years?
• RQ2: What are the structures of conversation and
information sharing networks in individual conferences?
Have these network structures changed over time?
• RQ3: Do users participate on Twitter for the same
conference over consecutive years? To what extent can we
predict users’ future conference participation?

12. This Paper: Twitter in Conferences over Time
Types of communication
over time
• RQ1: Do users use Twitter more for socializing with peers or
for information sharing during conferences? How has such
use of Twitter during conferences changed over the years?
• RQ2: What are the structures of conversation and
information sharing networks in individual conferences?
Have these network structures changed over time?
• RQ3: Do users participate on Twitter for the same
conference over consecutive years? To what extent can we
predict users’ future conference participation?

13. This Paper: Twitter in Conferences over Time
• RQ1: Do users use Twitter more for socializing with peers or
for information sharing during conferences? How has such
use of Twitter during conferences changed over the years?
• RQ2: What Network are the structures changes
of conversation and
information sharing networks in individual conferences?
Have these over network time
structures changed over time?
• RQ3: Do users participate on Twitter for the same
conference over consecutive years? To what extent can we
predict users’ future conference participation?

14. This Paper: Twitter in Conferences over Time
• RQ1: Do users use Twitter more for socializing with peers or
for information sharing during conferences? How has such
use of Twitter during conferences changed over the years?
• RQ2: What are the structures of conversation and
information sharing networks in individual conferences?
Have these network structures changed over time?
• RQ3: Do users participate on Twitter for the same
conference over consecutive years? To what extent can we
predict users’ future conference participation?
Prediction of
User Participation

15. RQ1: Types of Conversation over Time
Conference
Tweets (line)
vs.
Random Sample
(dashed)

16. RQ1: Types of Conversation over Time
Conference
Tweets:
- RT and URLs
increase over
time

17. RQ2: Network Changes over Time
Conversation Network: Mentions & Replies
vs Retweet Network

18. RQ2: Network Changes over Time

19. RQ2: Network Changes over Time
Weakly Connected Components:
Conversation network: get more scattered.
RT network stays the same over time.

20. RQ3: Prediction of User Participation
• Can we tell whether someone will return the coming year to the
conference based on her participation?
AUC Results
Baseline: Number of tweets in the User Timeline (with +
without hashtag during the conference weeks)
Your activity over the
conference week is a
good baseline predictor

21. RQ3: Prediction of User Participation
• Can we tell whether someone will return the coming year to the
conference based on her participation?
AUC Results
Baseline: Number of tweets in the User Timeline (with +
without hashtag during the conference weeks)
Only #hashtagged
tweets doesn’t help
much

22. RQ3: Prediction of User Participation
• Can we tell whether someone will return the coming year to the
conference based on her participation?
AUC Results
Baseline: Number of tweets in the User Timeline (with +
without hashtag during the conference weeks)
If you get retweeted by
many people, or by
important people

23. Conclusions & Future Work
• Usage of Twitter in Conferences has increase over time and has
changed considerable: people are more likely to RT and share URLs
than to “talk” (replies and mentions) by Twitter
• Over time, the conversations get more scattered, but not the
“advertisement” or “resource sharing” of the RT network
• Total Twitter activity over the conference week helps to predict
returning to the conference, not only #hashtagged Twitter activity.
• CURRENT EXTENSION: use content-based features, sentiment
analysis, consider hypotheses of “communities of practice”

24. Limitations
• The analyses only considered types of messages (re-tweets,
mentions, URLs) and social network analysis’ metrics.
• Obtaining a representative sample of Twitter is a problem not solve
(at least for researchers in academia that can use only the Twitter API,
or other 3rd party APIs such as Topsy)
• We are covering a sample of conferences in the domain of Computer
and Information Science.

25. A little spoiler…
Subjectivity: Does this
message have an opinion
is rather a fact?
 “I think this is a nice
talk” (subjective ++)
 There are 10 people
in this room (
subjective-- )

26. A little spoiler… #ht2014?
#ht2014 ?
Polarity
HT

27. Thanks!
• And don’t forget today the social event (is included with the
conference registration) at
• Hotel Plaza San Francisco, 8:30pm

28. … but then, I decided to give a step back
• Our first attempt was studying 4 conference-communities (Wen,
Parra, Trattner at CSCW 2014) and see how groups of conference
attendees communicate on Twitter
• Faculty / Senior research students / Junior research students
• Others
• We expected to have an important participation from newcomers to
the conferences.

29. How this Research Started
• It started by creating a tool, Conference Navigator, in order to
support conference attendees
Modify/Enhance
Communities
Behavior
• Conference Navigator provides a ranking of top bookmarked talks,
recommendation of talks, “People who bookmarked this, also
bookmarked …”

30. … but then, I decided to give a step back
• Our first attempt was studying 4 conference-communities (Wen,
Parra, Trattner at CSCW 2014) and see how groups of conference
attendees communicate on Twitter
• Faculty / Senior research students / Junior research students
• Others
• We expected to have an important participation from newcomers
during the conferences.
We found out that Twitter
might not be always the best
entrance for newcomers in a
research community.

31. Random Sampling
• Use of TOPSY API: Search for a –or- b –or- …. –z-
• Pick 2/3 of the hours in a year (24 x 365)
• Per each hour sampled, pick randomly 2 minutes
• As a result, there will be a lot of tweets in several pages. Pick
randomly a result page
• Total tweets sampled: ~ 5.8 million tweets (2009 -2013)
• Distribution of RT, replies, URLs is not significantly different than D.
Boyd, S. Golder, and G. Lotan, “Tweet, tweet, retweet: Conversational
aspects of retweeting on twitter,” in System Sciences (HICSS), 2010
43rd Hawaii InternationalConference on, pp. 1–10, IEEE, 2010.