a longitudinal analysis of blog linkages among south korean politicians han woo park, ph.d....

33
A longitudinal analysis of blog A longitudinal analysis of blog linkages among South Korean linkages among South Korean politicians politicians Han Woo PARK, Ph.D. Han Woo PARK, Ph.D. Assistant Professor Assistant Professor YeungNam University YeungNam University Korea (South) Korea (South) [email protected] [email protected] http://www.hanpark.net http://www.hanpark.net This is in collaboration with Randolph Kluver This is in collaboration with Randolph Kluver at Texas A&M University, USA at Texas A&M University, USA presented at the international conference of NetSci 2007, May 22-25, 2007, New York City, USA. http://www.nd.edu/%7Enetsci/index.html

Upload: beverley-marshall

Post on 26-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

A longitudinal analysis of blog A longitudinal analysis of blog linkages among South Korean linkages among South Korean

politicians politicians

Han Woo PARK, Ph.D.Han Woo PARK, Ph.D.Assistant ProfessorAssistant Professor

YeungNam UniversityYeungNam UniversityKorea (South)Korea (South)[email protected] [email protected]

http://www.hanpark.nethttp://www.hanpark.net

This is in collaboration with Randolph Kluver This is in collaboration with Randolph Kluver at Texas A&M University, USAat Texas A&M University, USA

presented at the international conference of NetSci 2007, May 22-25, 2007, New York City, USA. http://www.nd.edu/%7Enetsci/index.html

Introduction

The Internet has become an important communication channel for politicians because of the power of relational networking

This study examines the online network relationships among political actors on cyberspace over a period of time

Trends of previous research

One approach has been on the content features of blogs, as a form of expression

The second approach takes a more network-oriented method using link data

However, little research has done on what impact the changing network structures might reshape network properties at an aggregate level

Issues in the use of political blog

Cass Sunstein (2001, 2004)

- The process in which messages on cyberspace are exchanged and filtered brings about a ‘balkanization,’ or fragmentation, among people along interest lines or other socially defining characteristics

Early studies (Ackland, 2005; Adamic & Glance, 2005) on political blog have produced results that generally confirm Sunstein’s claims about the Internet’s fragmenting effect

Issues in the use of political blog

In contrast, Hargittai, Gallo, & Kane (2005)

- While polarization theory is useful in explaining blog-mediated social communication

- It was also true that cross-linking across ideological borderline was happening

- Also found that the divide did not tend to grow greater over time

Issues in the use of political blog

Gap in literature review

Most of the previous research has looked at blogs only at a specific point in time, with little attention paid to the evolution of the online networks

Another gap is that little research has done on blogs maintained by “politicians”

Research question

How do the network properties of the Korean politicians’ blogosphere change over time?

Specifically, does the network undergo any significant changes that would suggest a longer term impact upon mediated politics in South Korea?

Sample and data collection

17th National Assembly members The ruling Uri Party (Uri) The major opposition Grand Nat’l Party (GNP) Two minor opposition parties and others

- The Democratic Labor Party (DLP)

- The Democratic Party (DP)

- Others (e.g., a few independents)

Sample and data collection During early July of 2005 and early

November of 2006 respectively In 2005, 107 blog-owning members were

taken and 125 members in 2006 As supplemental data, we interviewed

politicians who have posted hyperlinks to at least one other Assembly member on their “blogrolls” (N=51)

Centrality: Individuals’ positions

The opposition GNP has a higher centrality ratio of its members and less isolates than the ruling Uri party in two years

In 2006, 21 out of 31 indegree isolates are from Uri party, and only six GNP isolates

The bloggers from the “minor” opposition also demonstrated poor network centrality

Compared to other minor players, half of

the DLP members maintained blogs and they are relatively well connected over time

Centrality: Individuals’ positions

Change in group properties

The indegree pattern was by and large consistent across years but the outdegree pattern shows a moderate change

The maximum value of indegree in 2006 is 26, down from 30 in 2005 (-13.3%)

In the case of outdegree, the maximum value of the year 2006 is much less than that of previous year (53 -> 41, -22.6%)

Change in group properties

The total number of blogging members has increased from 107 to 125

This implies that blog has been a part of general-purpose technologies for online communication

Distributions of degree centralities of 107 politician blogs in 2005

0

10

20

30

40

50

60

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106

Degree Rank Order

Deg

ree

Indegree Outdegree

Distributions of degree centralities of 125 politician blogs in 2006

0

10

20

30

40

50

60

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97101105109113117121125

Degree Rank Order

Deg

ree

Indegree Outdegree

Change in group properties

Although the number of blogging Assembly members stayed roughly consistent, the network has become sparser, less integrated, and decentralized over time

Network density decreased from 0.07 to 0.04, indicating a decline of 37.39 percent

Change in group properties

As the average number of hyperlinks connections has decreased and the number of isolates has increased, integration among members has been weakened

Centralization also showed much change overtime, as a few hub members are disappearing

Change in group properties These changing topographies reflect

different communication practices Politicians become less enthusiastic about

using blog in order to make a tie with others

Our interviews reinforced our finding that politicians are gradually turning away gradually from blogs to mini-hompies, or to put a priority on official websites

Changes from 2005 to 2006

Year Nodes Density Average # of links

Centralization (In)

Centralization (Out)

Isolates

2005 107 0.07 In7.5

Out12.3

21.7 43.6 In18

Out60

2006 125 0.04 6.4 9.6 16.9 29..0 31 74

Yearly change

Percent -42.9 -14.7 -21.9 -22.1 -33.5 72.2 23.3

Network diagram The map clearly partitions relatively

peripheral groups (e.g., a set of isolates) from central ones in South Korea

Interestingly, the two major parties, the ruling Uri Party and the opposition Grand National Party, had clearly delineated networks among their members

Block modeling over time The group density of 2006 for the ruling Uri

party network jumped down to 0.018 from 0.026 in 2005

Group density of the GNP also fell to 0.128 (2006) from 0.187 (2005)

This means that blogs did not strengthen a sense of being a ‘closed-network’ community over time

In relation to the deepening divide, there is only one cross-connection between Uri and GNP in 2005

In 2006, the number of connections from Uri to GNP had increased to 10 and the frequency of reverse linking (GNP -> Uri) had gone to 3 in 2006

The number of cross-links has slightly increased over time

Block modeling over time

N = 38Density = 0.026Link Sum = 36

N = 63Density = 0.187Link Sum = 731

N = 5Density = 0.2Link Sum = 4

Network diagram in 2005

N = 49Density = 0.018Link Sum = 42

N = 5Density = 0.2Link Sum = 4

N = 69Density =0.128Link Sum=599

Network diagram in 2006

A Scale-free network?

A characteristic of a scale-free network

- A small number of sites receive the majority of hyperlinks and most sites receive very few hyperlinks

- The number of hyperlinks per site has a power law distribution

- The top 1% of sites attracted more than half of the hyperlinks

Change in group properties We did not see the characteristic signature

of a power law distribution from inlink connectivity among South Korea’s politician blogs but there may be some characteristics from outlink distribution

The outlink distribution in log-log scale is weakly fitted to the tail of a power law distribution when compared to real tails

0

5

10

15

20

25

30

35

0 5 10 15 20 25 30 35

Number of i nl i nks

Numb

er o

f bl

ogs

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.5 1 1.5 2

Number of i nl i nks

Numb

er o

f bl

ogs

•Indegree distributions in linear scale plot in 2006 •Indegree distributions in log scale plot in 2006

Power law analysis in 2006A characteristic of a power law distribution? A small number of site receive the majority of links and most sites receive very few links

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60Number of out l i nks

Numb

er o

f bl

ogs

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8Number of out l i nks

Numb

er o

f bl

ogs

•Outdegree distributions in linear scale plot in 2006 • Outdegree distributions in log scale plot in 2006

Power law analysis in 2006

Thanks for your attention !Thanks for your attention !

Interesting & growing areaPromising approach

New Media & Society Lab at YeungNam University in S. Korea

* Website : http://www.hanpark.net * E-mail : parkhanw oo@ hotm ail.com

Many thanks to my assistants!