structural changes in_the_2003-2009_global_hyperlink_network_(ica)_2
TRANSCRIPT
Is global cyberspace becoming individualized or globalized?
Han Woo Park Department of Media and Information, WCU Webometrics Institute, YeungNam University, South Korea
George A. BarnettDepartment of Communication, University of California – Davis
Chung Joo ChungDepartment of Communication, State University of New York at Buffalo
Introduction
The widespread diffusion of the Internet & telecommunication networks
Global communication systems strengthen the worldwide social relations and cultures are shaped and clustered across national borders (Barnett & Sung, 2006)
Few studies have examined the international Internet’s structure
An alternative approach is the analysis of inter-domain hyperlinks (Barnett, et al, 2001).
Hyperlink is the technological capability that enables a website to link with another WWW is a distributed hypertext system consisting of a virtual network of content and
hyperlinks, with billions of inter-linked pages (Almind & Ingwersen, 1997; Kleinberg & Lawrence, 2001)
Web’s social structure as a technological link and communication network at the level of nation-states
Global Communication Networks The structure of international telecommunication
(1) Increased centralization of communication flows (Barnett et al., 1996; Barnett & Park, 2005; Chase-Dunn, 1989; Chase-Dunn & Grimes, 1995; Lee et al., 2007; Wallerstein, 1974)
rooted in a broader perspective of economics emphasizing the asymmetry between countries
world-system theory (Barnett et al., 1996; Barnett & Park, 2005; Chase-Dunn, 1989; Chase-Dunn & Grimes, 1995; Lee et al., 2007; Wallerstein, 1974)
(2) Increased diversification of communication flows (Barnett, 2001; Danowski, 2000; Lee et al., 2007; Matei, 2006; Monge & Matei, 2004; Robertson, 1992)
increasing trends of decentralization, regionalism or cultural pluralism
determination factors - countries’ geographical location (Barnett & Choi, 1995)
- languages (Barnett & Choi, 1995)
- religion (Barnett et al., 1999)
- national culture (Barnett & Sung, 2006)
Literature Review and Related Theories
Longitudinal Approach Longitudinal studies with respect to the Internet are difficult to conduct Some attempts to automatically index the changes within a website over
time Little research in examining the longitudinal hyperlink structure
generated by WWW on the international level Did global cyberspace follow the main trend toward globalization and
centralization of resources or become individualized and fragmented?
Literature Review and Related Theories
Research Questions
RQ1: What is the current (2009) structure of the international hyperlink network?
RQ2: How has the international hyperlink network changed between 2003 and 2009?
TLD Data Hyperlink connectivity between websites that belong to ccTLD AltaVista and Yahoo were used respectively for 2003 (Barnett & Park,
2005) and 2009 LexiURL Searcher
Automatic analysis of the impact of collections of web sites Link Impact Report
In the case of the U.S., three TLDs reserved for the exclusive use of American institutions, .edu, .gov, .mil were combined with .us
Methods- Data
The basic network data set an n x n matrix S, where n equals the number of nodes in the analysis The basic network data set is an n x n matrix S, where n equals the
number of nodes in the analysis Each cell, Sij , indicates the strength of the relationship among nodes i
and j
Freeman’s degree centrality, Bonacich’s eigenvector measure, hierarchical cluster analysis, MDS, Gini coefficient and QAP correlation and regression
Methods- Network analysis
Results
Table 1 International Hyperlink Centrality of the Top 30 Countries
Results
Figure 1. International Hyperlink Flow Network (2009)
The size of the concentric circles indicates the hyperlink connection density among countries. The thickness of the line connecting two nodes is proportional to the connection density between the two nodes. Only those ties exhibiting greater than 1 million hyperlinks are shown. All isolates have been removed from the figure. N=251.
Results
Figure 2. Incoming International Hyperlink Drawn using ManyEyes.com
Results
Table 2. International Hyperlink Networks Comparison (2003 vs 2009)
Results
Figure 3. International Hyperlink Structure in 2003
Only those ties exhibiting greater than 50,000 hyperlinks are shown (Barnett & Park, 2005). N= 47.
Results
Figure 4. International Hyperlink Structure in 2009
Only those ties exhibiting greater than 400,000 hyperlinks (eight times more compared to 2003, based on eight times degree difference) are shown. N=47.
Results
QAP Correlation & Regression The relation between the centrality of the 2003 and 2009 hyperlink
networks of the 47 countries was significant (in-degree, r = .798, p < .01; out-degree, r = .815, p < .01; eigenvalue, r = .821, p < .01).
The QAP correlation between the 2009 and 2003 data was only .406 (p < .01), accounting for only approximately 16% of the variance in the two networks.
Results
Some obvious and interesting changes The international hyperlink network became more centralized over time.
The composite score (Gini-based core/peripheral measure) of the 2009 network was 0.466, whereas it was only 0.291 for the 2003 network.
The greatest changes were found among the most central countries. Europe as a whole was much more central, particularly Germany.
The BRIC (Brazil, Russia, India and China) countries showed a number of changes.
The numbers of Brazil’s and Russia’s hyperlinks increased from 2003 to 2009. China had fewer outward links. India had fewer inward links than expected.
The centralities were distributed as a power curve (R2 ranging from .798 to .821) as Barabasi (2002) predicted, suggesting disproportional growth in the number of hyperlinks by the increased nodes (countries).
Whereas there was only one group in 2003, regional, cultural, and linguistic groupings formed in Latin America, Scandinavia, and around China and Russia in 2009.
Discussion
The structural position of a country is a result of its interactions with other countries
The structural position of a country determines its potential for development and its interaction patterns
The international hyperlink network became more centralized around G7 nations and Spain, with Western European countries joining the USA at the core.
At the same time, regional clusters developed in what may be characterized as the semi-periphery.
Otherwise, the overall structure of the international hyperlink network remained relatively stable.
Limitation
No indicators of the reliability of the measurement procedures or the validity of Yahoo’s search engine
Difficulties in addressing the nodes of Canada, Germany, and Indonesia, which share their domain names with various U.S. states
.tv, Tuvalu’s domain name, is primarily used by the television (‘TV’) / entertainment industry
gTLDs such as .com, .net, .org or .eu websites need to be decomposed
Thank you for listening