New directions for blog network mapping [with Lars Kirchhoff and Thomas Nicolai, IR11, Gothenburg, October 2010]
New directions for
blog network mapping
IR11
Göteborg, Sweden, 22 October 2010
Tim Highfield
Queensland University of Technology
Lars Kirchhoff
Thomas Nicolai
Sociomantic Labs
Overview
Blog network mapping
visualising connections between groups of sites
links, affiliations,
friendships, clusters
references, opposition
for large datasets, enables depiction
of non-physical space, make sense of data,
(although risk of focussing on 'sciencey'
visualisations rather than analysis)
Previous examples
Adamic and Glance (2005)
U.S. - 2004 Presidential election
RTGI/linkfluence (2007 - present)
France – 2007 Presidential election
U.S. - 2008 Presidential election
Europe – 2009 EU elections
Kelly and Etling (2008)
Persian-language
Etling et al. (2009)
Arabic blogs
Strengths
Overview of large/long-term datasets
Identification of clusters/groups, including around themes
or activities
Locate most active sites, popular sources
Weaknesses
Differentiation between types of link, link semantics,
link frequency
Temporal variations – event-oriented peaks or troughs,
context for blog activity
Research context
Study of French and Australian political
blogs (and associated sites)
Data collection: January – August 2009
Data used: blog posts and links
Focus on comparing whole period activity
and event- based, topical networks
New directions
1. Types of link
Blogrolls vs. in-text citations
(vs. in-text in topical network)
(vs. comments, other platforms)
2. Dynamics / Snapshots
Isolating posts around particular topics/
people/events
Day-by-day variations,
changes in network over time
Topical references vs. most cited resources overall,
topical blog activity compared to total output during period
Topical network: Obama inauguration
French political blogs
16-25 January 2009
Size:
in-degree
Colour:
out-degree
3. Mixed methods
Hyperlink analysis + content analysis
locating networks using link data from posts
discussing particular topic (+ visualisation)
analysing text of posts from individual blogs
to identify prominent themes,
different approaches to the same topic
4. Geographical representations
Distribution of sites within a link-based network
compared with geographical distribution of blogs
clustering around major cities?
blogging demographics? (urban/rural?)
Combined with content analysis – different topics being
discussed around regions?
5. Beyond the network paradigm?
Hyperlink analysis invites the use of 'networks'...
Other mapping approaches available, depicting
prominence of sites (traffic, in-links, rankings),
thematic clusters,
location, ideologies...
Druaux (2007)
Cartographie diablement subjective et approximative
de la blogarchie francophone 2007
http://www.ouinon.net/index.php?2007/09/24/215-cartograhie-blogosphere-
francophone