Cross-Platform Profiling tutorial at the Digital Methods Summer School 2013
Workshop Digital Methods Summer School 2013
University of Amsterdam/Goldsmiths, University of London
What is profiling?
• Specific approach in issue mapping.
• Issue mapping as the study of topical affairs.
• Cross-platform profiling studies issues and their
variation across multiple online spaces.
• In which platforms do issues occur?
• What actor composition?
• What thematic framing?
• Pace, rhythm, variation over time.
Profiling & Issue Mapping
• Studying topical affairs.
• Issue mapping, or controversy analysis, has been
developed as a research method in the field of
Science, Technology and Society (STS). (Callon,
1986; Barry, 2001; Latour, 2007)
• Empirical, processual approach.
• Asks: Is this topic an issue? Who are the actors?
Where is it based? Where is the issue happening?
How does it change?
Issue Profiling Online
• Digitization offers opportunities for
issue/controversy analysis (Rogers &
Marres 2000, Latour et. al 2007,
2010; Yaneva 2007):
• Explosion of digital traces and
analytical devices deploying
• A special focus on taking advantage of
medium-specificity for profiling.
Profiling & specificity
• Interested in the specific articulation of issues
within and across different platforms.
• What are the specific forms of participation per
• Starting point: medium- and platform specifity.
• Deploy the pre-structured character of platforms
for analytical purposes (Marres & Weltevrede
• Media vs. issue dynamics.
Taking grammars of action into
• Which grammars of action/data entities (Agre
1994) are relevant for profiling?
• Google: Search results, hosts, titles.
• Twitter: Queries, hashtags, users, mentions.
• Facebook: Groups, Pages, Likes, Posts, user
• Flickr: Pictures, tags, groups.
• Wikipedia: Articles, related articles, edits, users…
Standardised activities & data forms cater to a
multiplicity of use practices.
1. Actor composition.
2. Key platforms.
3. Bias and leaning.
4. Issue framing.
5. Variation over time.
6. Media effects.
7. Social dyanmics.
Case 1 Cross-platform
profiling of Fukushima
• How is the issue of the Fukushima
nuclear disaster discussed in the web,
blog and news sphere?
• Word frequency analysis of Google
• Do comments offer a different framing
of the issue?
1. Query Fukushima ~nuclear
2. Take top 25 URLs from Google Web,
Google Blogs, Google News
3. Copy paste content from URLs
4. Create a tagcloud
• Scraping as systematic extraction
of pre-formated data.
• Insights: different framing per sphere,
which sub-issues are trending right now.
• Q1 What about variation over time?
• Q2 Are there other measures beyond
Case 2 Variation over time
• Currency (liveness): Which aspects of
issues are peaking/trending at the
• Currency: word frequency analysis
• Variation over time (liveliness): changes
in association & framing of issues?
• Variation: Co-occurrence & co-word
• Network analysis of textual data: tracing
relations between terms based on ‘co-
occurrence’ (Callon et al. 1983,
Case 2 Profiling Crisis
• Interest in topical framing of crisis/austerity
across platforms (Marres & Weltevrede
• Mapping the liveness (currency) vs. liveliness
(variation over time) of “crisis” in Google and
• Early 2012.
• Google: Take search result titles into account.
• Twitter: Take hashtags into account.
Case 3 Profiling Hashtags
• Intra-platform profiling of hashtags.
• Starting point: Hashtags as specific grammar
of action to demarcate conversations.
• How can we further profile hashtags and give
account to their liveliness?
• Case: Twitter data on Climate Change.
• Period: 01.02. - 15.06.2012, 204795 tweets.
• WCIT conference in Dubai, 3-14
Dec 2012 organised by the ITU.
• Interval: 23.11 – 19.12.
• 108781 Tweets.
• Before data capture: Collection of
NGO and issue expert hashtags.
• Question: How does the issue vary
over time and whose terms are
being taken up or not?
• Part of the Digital Methods Toolkit.
• Name: dmi
• Password: twitter
• Campaign Hashtags: #netfreedom,
• Institutional Hashtags: #isoc,
• Campaign issues dominated at the
beginning, institutional ones at the
Before the conference
• Pushed before the conference,
weaker associations during.
Start of the conference
• Campaign hashtag.
End of the conference
• Institutional hashtag.
• Expert hashtag.
• Very diverse profile.
• Expert hashtag.
• Declining presence and less
• How does media liveliness map
into issue liveliness? How is
variation defined by grammars of
• Media-liveliness: bursty hashtags,
• Can we perceive medium- and
issue-specificity as a spectrum?
• Further research: How can we
profile social dynamics?
• What makes an issue more or less
• Go to DMI TCAT tool.
• Select “prism” dataset.
• Adjust the time interval to one
• Open Associational Profile in
• Create profile for term
“prism” or “spying”.
• Interval: Weekly.
• Exclude: prism, snowden.
• Which words to stand out?
Create further profiles for