1) The document analyzes Twitter data related to the CMAD2013 conference to explore how visual network analysis of social media data can provide insights for co-organizing conferences.
2) It describes creating a two-mode network of Twitter users and hashtags as well as one-mode networks of connections between users and hashtags. This revealed the most influential users, topics of common interest, and presented concepts.
3) The network analyses helped identify popular content and people with similar interests to better design future conferences, and highlighted how hashtag discussions can be facilitated or risk scattering. It also noted limitations of broken links in Twitter data over time.
Visual network analysis of Twitter data for co-organizing conferences: Case CMAD2013
1. Visual network analysis of
Twitter data for co-organizing
conferences: Case CMAD 2013
Jari Jussila1, Jukka Huhtamäki1, Kaisa Henttonen2,
Hannu Kärkkäinen1, Kaisa Still3
1 Tampere
University of Technology
2 Lappeenranta University of Technology
3 VTT Technical Research Centre of Finland
2. Overview of the study
• The aim of this research was to explore
what kinds of insights information
visualization of social media data can provide
for co-organizing conferences
• Case study
– The CMAD2013 (Community Manager
Appreciation Day) event held during 28 January
2013 in Finland
– 155 people participated in CMAD2013 event, and
223 people to the live stream during the day
– We collected a total of 2686 tweets over a six-week
period starting from January 21, 2013. 2138
tweets were exchanged during the day
– On the average, one participant sent more than 5
tweets during the day, even if also the online users
were counted in
2
3. Visual network analysis
•
•
Network analysis introduces a set of methods, practices and metrics for
supporting the investigation and representation of social media data.
We applied the following visual network analysis:
1. A two-mode network including two types of nodes, representing both
Twitter users and hashtags. Where a pair of users is connected to each
other when one has mentioned the other. Users are also connected to
the hashtags they have used in their tweets as well as to the hashtags
that are used in the tweets they have been mentioned in
2. One mode network of interconnections between people communicating
over Twitter that is users mentioning each other in tweets through
commenting, discussions and retweets
3. One mode network of co-occurence of hashtags included in the tweets
•
We complemented the network analyses with temporal analysis
through timeline views. For this, we followed an approach that Ebner
and Reinhardt (2009) introduced in which two splines are used to show
the cumulation of tweets.
Reference: Ebner, M. and Reinhardt, W. Social networking in scientific conferences–Twitter as tool for
strengthen a scientific community. Proceedings of the 1st International Workshop1.2.2014
on Science, (2009).3
4. Cumulation of the collected
tweets before, during and after
CMAD 2013
1.2.2014
4
5. Snapshot of two-mode network of people tweeting and their discussion
topics before, during and after the conference day. Interactive version is
available: http://www.tut.fi/novi/hicss2014/
1.2.2014
5
6. Network of people tweeting
1. the most influential people
2. people with similar interests
3. most interesting presentations
1.2.2014
6
7. Network of hashtags
1. most discussed topics
- cmadfilabel
- sketchnotes
- streaming
- swarm
2. most interesting
concepts and presentations
1.2.2014
7
8. Discussion and conclusions
• Insights of network of people
– Help to identify most interesting content for planning future conferences, e.g.
which tracks and topics
– Help to identify people with similar interests and for example plan sessions or
networking events that interest certain groups of people
• Insights of Twitter hashtag networks
– Provide implications how the discussion could be better designed and
facilitated, for instance, discussions tend to scatter when hashtags are
created bottom-up
– Also bring forward that Twitter data where the content links were created bottomup, e.g. by the conference participants, in some cases led to broken links or
discontinued services and thus missing data. Missing data in the sense that the
content of the link is no longer available, and as a consequence does not
accumulate to the knowledge base of past and future conferences
8
9. DOWNLOAD
http://urn.fi/URN:NBN:fi:tty-201401221053
CITATION
Jussila, Jari; Huhtamäki, Jukka; Henttonen, Kaisa; Kärkkäinen, Hannu;
Still, Kaisa 2014. Visual network analysis of Twitter data for coorganizing conferences: case CMAD 2013. Proceedings of the 47th
Annual Hawaii International Conference on System Sciences, January
6-9, 2014, Computer Society Press, 2014, 1474-1483.
ACKNOWLEDGMENTS
This research is sponsored by Tekes – the Finnish Funding
Agency for Technology and Innovation (Projects “Soila”; Innovative
Value Creation and Business Models of Social Media in B2B Networks,
and “Reino”; Relational Capital for Innovative Growth Companies).
1.2.2014
9