Mapping the Australian Networked Public Sphere
Axel Bruns, Jean Burgess, Tim Highfield ARC Centre of Excellence for Creative Industries and Innovation, Brisbane, Australia
Lars Kirchhoff, Thomas NicolaiSociomantic Labs, Berlin, Germany
Image by campoalto
Mapping Blog Networks
•Standard methodology:• find blogs (search, Technorati, specific blog platform, etc.)• identify links (on current page) crawl to linked pages repeat• capture (scrape) text and other details (not always included)• plot link network structure, correlate with blog content patterns
•Problems in blog mapping:• defining and identifying the population to be mapped• determining which links are relevant• method of plotting links, identifying blog clusters, etc.• correlating link network structure and blog themes• tracking changes over time
Key Problems
• Technology limitations:• crawlers and scrapers often lack sophistication
• need to distinguish: • posts – comments – ancillary / functional texts
• discursive links – blogroll links – functional links
• want to slice data in different ways:• select blog activity for specific days, weeks, months
• select blog content and links for specific blogs or blog clusters
• Analytical limitations:• patterns of interlinkage tell only part of the story
• maps provide only a temporary snapshot
• want to understand:• what clusters have in common
• and how they change over time
Our Approach
• Process stages (Australian political blogs as test case):• data gathering and processing
• track large number of (broadly) political Australian blogs through RSS feeds• scrape blog content for newly posted entries• separate blog post content from ancillary materials / separate discursive links from other link types• (grow master list of blogs as required)
• content analysis• combine extracted blog post content (per blog, per cluster, per timeframe, …) • automated analysis to identify key themes and keywords • currently using Leximancer
• network analysis• combine extracted link information (overall, per timeframe, per cluster, …)• automated network mapping to identify lead blogs and clusters• currently using Gephi
• combined analysis• e.g. comparative content analysis for lead blogs and clusters in the link network• e.g. correlation of blogosphere patterns with external factors (parallel themes in mainstream media,
etc.)
Patterns of Activity (Jan.-Aug. 2009)
0
100
200
300
400
500
600
700
2009
.01.
1220
09.0
1.14
2009
.01.
1620
09.0
1.18
2009
.01.
2020
09.0
1.22
2009
.01.
2420
09.0
1.26
2009
.01.
2820
09.0
1.30
2009
.02.
0120
09.0
2.03
2009
.02.
0520
09.0
2.07
2009
.02.
0920
09.0
2.11
2009
.02.
1320
09.0
2.15
2009
.02.
1720
09.0
2.19
2009
.02.
2120
09.0
2.23
2009
.02.
2520
09.0
2.27
2009
.03.
0120
09.0
3.03
2009
.03.
0520
09.0
3.07
2009
.03.
0920
09.0
3.11
2009
.03.
1320
09.0
3.15
2009
.03.
1720
09.0
3.19
2009
.03.
2120
09.0
3.23
2009
.03.
2520
09.0
3.27
2009
.03.
2920
09.0
3.31
2009
.04.
0220
09.0
4.04
2009
.04.
0620
09.0
4.08
2009
.04.
1020
09.0
4.12
2009
.04.
1420
09.0
4.16
2009
.04.
1820
09.0
4.20
2009
.04.
2220
09.0
4.24
2009
.04.
2620
09.0
4.28
2009
.04.
3020
09.0
5.02
2009
.05.
0420
09.0
5.06
2009
.05.
0820
09.0
5.10
2009
.05.
1220
09.0
5.14
2009
.05.
1620
09.0
5.18
2009
.05.
2020
09.0
5.22
2009
.05.
2420
09.0
5.26
2009
.05.
2820
09.0
5.30
2009
.06.
0120
09.0
6.03
2009
.06.
0520
09.0
6.07
2009
.06.
0920
09.0
6.11
2009
.06.
1320
09.0
6.15
2009
.06.
1720
09.0
6.19
2009
.06.
2120
09.0
6.23
2009
.06.
2520
09.0
6.27
2009
.06.
2920
09.0
7.01
2009
.07.
0320
09.0
7.05
2009
.07.
0720
09.0
7.09
2009
.07.
1120
09.0
7.13
2009
.07.
1520
09.0
7.17
2009
.07.
1920
09.0
7.21
2009
.07.
2320
09.0
7.25
2009
.07.
2720
09.0
7.29
2009
.07.
3120
09.0
8.02
2009
.08.
0420
09.0
8.06
2009
.08.
0820
09.0
8.10
2009
.08.
20
Opinion
Blog
Australian News
Bushfires Budget
Artefact
Qld Election
New Sites Added
Content Analysis: Individual Blogs
Content Analysis: Political Blogosphere
4-5 August 2009
Next Steps
•ARC Discovery Project, 2010-12 (Axel Bruns and Jean Burgess)
•Expand the population:• From political blogs to Australian blogs in general• Snowball / network crawler approach
• Map the blogosphere:• Long-term map of frequent interlinkages, clusters of blog communities• Short-term maps of ad hoc networks around current events and themes• Trending topics and correlations with mainstream media• Patterns of information flow, structures of dissemination and influence
• Extend to other online publics:• Twitter, Flickr, YouTube – what can we track, what publics do we find?• How are these spaces interconnected?