1. Tracking Social Media Participation:
New Approaches to Studying
Dr Axel Bruns
ARC Centre of Excellence for Creative Industries and Innovation
Queensland University of Technology
firstname.lastname@example.org – @snurb_dot_info
http://snurb.info/ – http://mappingonlinepublics.net/
Image by campoalto
2. Researching Social Media
• Social Media:
Websites which build on Web 2.0 technologies to provide space for in-depth
social interaction, community formation, and the tackling of collaborative
Axel Bruns and Mark Bahnisch. "Social Drivers behind Growing Consumer Participation in User-
Led Content Generation: Volume 1 - State of the Art." Sydney: Smart Services CRC, 2009.
3. Researching Social Media
• Various existing research approaches:
• Processes and practices How? What?
• Content generated by users What?
• Sites and organisational structures How? In what context?
• User surveys (demographics, practices, motivations) Who? Why?
• Content coding (usually small-scale) What?
• Mostly small-scale – limited applicability?
4. Known (Un)knowns
• What we know:
• Behaviour of small social media communities
• Practices of lead users
• Structural frameworks for selected sites / site genres
• Broad demographics of social media users
• Some things we want to know:
• How does all of this work at scale?
• What about ‘average’ users?
• How do communities overlap / interact?
• Can we track developments over time?
5. (Kelly & Etling, 2009)
6. Mining and Mapping
• New research materials:
• Massive amounts of data and metadata generated by social media
• Mostly freely available online (Web / RSS / API access)
• Often in clear, standardised formats
• New research tools:
• Network crawlers (e.g. IssueCrawler)
• Website scrapers / capture tools (e.g. Twapperkeeper)
• Network analysers / visualisers (e.g. Gephi, Pajek)
• Large-scale text analysers (e.g. WordStat, Leximancer)
7. Asking Sophisticated Questions
• What timeframe?
• Crawler approach: anything posted in the last 20 years
• Resulting in one static map – but what’s happening now?
• What map?
• Other ways to categorise these sites?
• Differences in activity, consistency
• Known unknowns – dynamics in the Iranian blogosphere:
• Sites appearing / disappearing?
• Increased / decreased activity?
• New linkage patterns:
• Stronger / weaker clustering?
• Move from one cluster to another?
• Change in topics, shift in emphasis, spread of information?
8. Asking Sophisticated Questions
• Problems with current research approaches:
• Crawlers don’t distinguish site genres or link types
• Scrapers gather all text (including headers, footers, comments, …)
• Very few attempts to trace the dynamics of participation
• Many different ways to visualise these data
• Assumptions often built into the software, and difficult to change
• Alternative approaches:
• Gather large population of RSS feeds (and keep growing it)
• Track for new posts, and scrape posts only (retain timestamp)
• Extract links and keywords for further analysis
• Develop ways of identifying and visualising change over time
• Needs to be appropriate to research questions
9. Applications: Twitter
• Who tweets, and what about?
• How do themes and topics
change over time?
• How do #hashtags emerge?
• What do users share – in
links and retweets?
• How do MSM stories influence
• How do follower networks and
10. #ausvotes on Twitter (17 July-24 Aug. 2010)
11. #ausvotes: Mentions of the Party Leaders
12. #ausvotes: Keyword Co-Occurrence
13. #ausvotes: Key Election Themes
14. Applications: Blogosphere
• (How) does the ‘A-List’
change over time?
• (How) does political
alignment change over time?
• How strong is cross-
connection across clusters?
• What topics are discussed
– e.g. compared with MSM?
• What happens when power (Adamic & Glance, 2005)
changes hands – is blogging
an oppositional practice?
• Beyond left and right (beyond politics!): identification of
blog genres based on textual / linkage patterns
(qualitative follow-up necessary)
15. Applications: Australian Blogosphere (partial)
16. Applications: last.fm vs. Billboard
• Tracking listening patterns:
• Billboard = sales charts
• last.fm = listening activity
• Comparing sales and use
of new releases
• Identifying brief flashes and
• Distinguishing casual listeners
and committed fan groups
• Providing market information
to the music industry
(Adjei & Holland-Cunz, 2008)
17. Application: Wikipedia Content Dynamics
• Tracking editing patterns:
• Identifying stable/unstable content
• Highlighting controversy, vandalism,
• Tracking consensus development
• Tracking responses to developing
• Establishing trustworthiness based (http://trust.cse.ucsc.edu/)
on extent of peer review
• Highlighting most hotly debated
(edited) sections of text
18. For More Ideas: VisualComplexity.com
19. _______ Science Emerges
• Web Science Research Initiative (Tim Berners-Lee et al.)
• Science, technology, computer engineering, …
• Limited inclusion of media, cultural, and communication studies
• Strong focus on Semantic Web, artificial ontologies
• Cultural Science + Cultural Science Journal (John Hartley et al.)
• Media & cultural studies, evolutionary economics, anthropology, …
• Limited inclusion of computer sciences, technology
• Strong focus on culture, innovation, evolutionary dynamics
• Data mining and visualisation
• Substantial commercial work on data mining
• Visualisation experiments in communication
design and visual arts
20. Looking Ahead
• Critical, interdisciplinary approaches
• Need to better connect cultural studies, computer science, research
• Need to interrogate in-built assumptions of existing technologies
• Need to explore and investigate visualisation and analysis methods
• Need to develop cross-platform approaches and connect with more
• Open questions
• Ethics of working with technically public, but notionally private data
• Potential (ab)use of data mining techniques and/or research results by
corporate and government interests
• What new knowledge can such research contribute?