America Is the Target; Israel Is the Front Line _ Andy Blumenthal _ The Blogs...
4.11 Korea General Election, Oxford Internet Institute Big Data Conference, Webecology, Webometrics
1. An analysis of web ecology examined
through social webometrics during 2012
general election in south Korea
Yoonjae Nam, PhD : Kyunghee University
Yeon-Ok Lee, PhD : Royal Holloway, University of
London
Han Woo Park, PhD : Yeungnam University
Internet, Politics, Policy 2012: Big Data, Big
Challenges?
Organised by the Oxford Internet Institute (OII,
University of Oxford) on behalf of the OII-edited
Web Ecology - 2012 journal Policy and Internet, the conference
academic 9/21/2012
will take place over two days (20-21 September 2012)
2. Outlines
Web ecology
Inter-relationship among websites by the
human activity of using the Internet in
information ecology
Web Public opinion &
Campaign
ecology Issues
Observing integration and changes of
diverse information behavior across four
platforms (News, Blog, Twitter, Facebook)
during the campaign period of the 2012
Web Ecology - 2012 9/21/2012
national elections in South Korea
3. Elections and Internet Use
Large amounts of information /Updated quickly
Opportunity to set their own agenda
Multimedia, More interactive, & Less dependent
on the journalist-controlled traditional media
Mobilizing voters, Encouraging donations, &
Enabling immediate transactions and recruitment
Hyperlink system
Allowing users to generate content -Public
participation
Interactive feedback functions, Discussion boards
& SNSs.
Web Ecology - 2012 9/21/2012
4. Web ecology and elections - 2
Information ecology is “a science which studies the
laws governing the influence of information summary on the
formation and functioning of bio-systems, including that of
individuals, human communities and humanity in general and on
the health and psychological, physical and social well-being of
the human being; and which undertakes to develop
methodologies to improve the information environment.” -
Eryomin (1998)
Integration of diverse types of
information
Recognition of evolutionary change
Observation and Description
Web Ecology - 2012 9/21/2012
Focus on people and information
5. Background
General election in Korea (April, 11, 2012)
Party (7) : Saenuri-dang (ruling party), Minju-dang
(major opposition party) and 5 minors.
Leader of party (9) : some party has joint leader
Guenhye Park, Myungsook Han, Junghee Lee and 6
leaders
Web Ecology - 2012 9/21/2012
6. RQs
RQ1: Can current political situation during
election campaigns be described effectively by
examining web ecology (big data analysis)?
RQ2: How differently do media platform serve to
preserve a web ecology during political election
campaigns?
Web Ecology - 2012 9/21/2012
7. Methods : Networked visibility of a
pair of actors
Networked visibility of a pair of actors
Networked visibility of a pair of actors: 190 queries
A contextual word, “election”, is inserted
Search query: Actor A, Actor B, election
Date option was used: 23 March 2012 when the NEC
closed candidate registration
Politica Politica
l party Election l party
A B
Web Ecology - 2012 9/21/2012
8. Methods: Networked visibility of a
pair of actors
Data collected from 4 April
Naver: Webpage, Blog, News article
Google API: Twitter, Facebook
20 actors: eight political parties and their party
leaders that nominated more than 20 candidates
for congressional election
Web Ecology - 2012 9/21/2012
9. Result 1
a series of network representation of the election
from each platform at twelve points (days) (April 4
to 15, 2012)
Network centralization scores which refer to what
extent a web-ecology from each platform would
be politically biased.
Results indicated that Twitter had the highest
centralization scores at all twelve points.
During twelve days, network centralization scores
of news and blog were relatively higher than
those of social media.
Web Ecology - 2012 9/21/2012
10. Results
• Social media (Twit, FB) show higher centralized structure of
network than news or blog media.
• Social media are more biased media in terms of political
communication than news or blog media.
News Blog Twitter FB
April 4 16.11% 15.81% 21.82% 21.45%
April 5 15.61% 15.62% 21.82% 21.45%
April 6 15.39% 15.74% 21.80% 20.72%
April 7 15.36% 15.74% 21.71% 20.33%
April 8 15.21% 15.82% 22.01% 20.34%
April 9 15.06% 15.87% 22.17% 20.71%
April 10 14.99% 16.01% 22.41% 20.68%
April 11 14.88% 16.02% 22.23% 20.80%
April 12 14.82% 16.24% 21.96% 19.67%
April 13 14.84% 16.34% 21.65% 18.91%
April 14 14.85% 16.32% 21.38% 18.14%
Web Ecology - 2012
April 15 14.86% 16.35% 21.09% 9/21/2012
17.93%
11. Results
Analyzing network centralities in four platform,
both major ruling and opposition parties (e.g.
Saenuri party and Minju party) and leaders of
those parties (e.g. Park guenhye and Han
myungsook) had generally higher degree
centrality scores than minor parties and leaders
of those parties.
However, results showed that there were
distinctive features and trends across four
platforms.
Web Ecology - 2012 9/21/2012
12. Results
In news platform, both Saenuri party and Minju
party have almost similar patterns of degree
centralities and the highest web visibilities.
In blog case, the opposition parties, Minju party
and UPP, received relatively more attentions from
bloggers as results showed the degree centrality
scores of those parties had higher than those of
the ruling party.
Web Ecology - 2012 9/21/2012
13. Results
Although, in Twitter case, the trend of degree
centrality during twelve days was similar to that of
degree centralities in news and blog cases, a
distinguished feature was a leader of UPP, Lee
junghee, who is one of power Twitter user (has
over twenty thousand followers) had more degree
centralities than a leader of Han, myungsook who
is the leader of major opposition party.
Degree centrality trends in Facebook case were
quite different comparing to others. Before
election day the trends of centrality were stable,
then the scores of centrality rapidly increased
Web Ecology - 2012 9/21/2012
after election.
14. Results
Correlations among centralities of 4 platforms
MinJu Party News Blog Twitter Facebook
- .976** .984** .644*
- .949** .594*
- .731**
Web Ecology - 2012 9/21/2012
15. Results
Linear trends of centrality
Linear Model Summary Parameter Estimates
Equatio R
n square F df1 df2 Sig. Constant b1
News .97 306.48 1 10 .00 23590.99 3338.03
Blog .87 65.48 1 10 .00 53217.77 2250.88
Twitter .97 318.22 1 10 .00 27325.92 1833.68
Facebo .48 9.49 1 10 .012 1733.05 120.72
ok
Centrality trend of Facebook: R square and Slope (b1) scores were relatively lo
Web Ecology - 2012 9/21/2012
18. Discussion
Networked visibility from big data is showing inter-
relationships between current political situation
and the data.
Both major parties and their had generally higher
degree centrality scores.
There were distinctive features and trends across
four platforms.
Social media (Twitter, Facebook) show higher
centralized structure of network than news or blog
media. Social media are more biased media in
terms of political communication than news or
blog media.
Web Ecology - 2012 9/21/2012
19. Discussion
Linearly increasing trend of centralities show
information concentration on specific parties and
leaders during campaign
After election, centralities increased very rapidly.
Facebook is relatively stable platform.
The information retrieval methods tracing the
trend of information seeking behaviors do not
require contacting voters individually
a complementary tool for polls.
Web Ecology - 2012 9/21/2012