본문 바로가기
Book and Article

The Structure and Dynamics of the Korean Twitter Network, Dukjin Chang & Ghi-Hoon Ghim(2011)

by SNU Sociology 2020. 4. 7.

Author
Dukjin Chang (Professor, Dep. of Sociology, Seoul National University)
Ghi-Hoon Ghim (CEO, Cyram inc./ Adjunct Professor, Dep. of Sociology, Seoul National University)

Source Journal of Communication Research. No.48(1). 2011 (link)

© natanaelginting - stock.adobe.com

This paper aims to grasp the structure and dynamics of Twitter through the analysis on Korean Twitter Network. The authors identified 1,133,365 presumably Korean Twitter accounts using snow-balling sampling method in early 2010. They then collected and analyzed all the tweets, Follow Relations, RT Relations, Reply Relations, and url cited by all those Korean Twitter accounts for two months from Aug. 1 to Sept. 30, 2010. As a result, 77,452,090 tweets in total were collected and it was found that out of 1.13 million twitter accounts, 761,177 accounts had written at least one tweet out of those tweets.

Based on this data, the paper shows findings in two parts. First, the paper introduces (1) the general characteristics of Twitter networks, explains (2) the characteristics of tweets that are often RT-ed, and (3) the characteristics of url that are linked to tweets. Second, to analyze the process of information spreading and arguing on Twitter, authors track two representative cases in which a lot of information/opinions were spread through Twitter during two months - (1) the heavy rain during Chuseok period (Breaking News case) and (2) 'ideological consumption' debate (Discussions).

As this was the first ever case of a sociological analysis on the entire Korean Twitter network, the authors insisted that it was meaningful to take the first step of exploration because the theoretical approach to Twitter networks was possible only when these technical and descriptive findings were accumulated to some extent. From fruifully exploratory analysis, there were many new findings on Twitter Network and authors could infer important implications for social science from them.

 

Common characteristics of the entire Korean Twitter network

© Adobe stock

1) General Characteristics of Korean Twitter Network

Based on analysis on more than 77 million tweets, authors proved that Twitter is basically a space for communication and discourse rather than mere personal/private journaling space. This was shown by the fact that three-quarters of all tweets are made up of RT or Reply. Twitter thus meets the basic requirements for social agenda setting and discussion functions.

Although the characteristics of the power-law distribution are well known and the Follow network of Twitter is also seen as following the power-law distribution, sociological interpretation on that requires a cautious approach. Just because a small number of users follow the majority does not necessarily indicate a monopoly of opinion. In particular, considering that about 4 out of 10 people can spread their opinions by being RTed by others, it is difficult to conclude the social consequences of distribution characteristics easily.

2) Which Tweets get Retweeted(RT)?

Authors filtered 10,000 most-RTed tweets and categorized them into 7 different fields ( Life, Politics, Social issues, Culture, IT & Science, Economy&Business, Sports ) and 3 different characteristics of content( Opinion & Emotion(Subjective)/ Information & News(Objective)/ AD & PR & Campaign). Analysis on the top 12 most RT-ed tweets in the political and social issues fields shows that there is an overwhelming amount of government criticism. This seems to be related to the alternative media nature of new media such as Twitter. As the conservative monopoly on offline traditional media gots worse during the time(2010), people tend to seek different opinions or information through New Media- a.k.a Twitter and other SNSs.

3) How is URL used?

Out of over 77 million tweets, 13.7 percent contained url, and about half of all users have linked url at least once. URLs are often Pictures or News Media (Newspaper, Broadcasts, Web news.. etc) The paper explains that this is due to the nature of Twitter, which limits users to write in only 140 characters. URL links are often needed in Twitter, resulting in a discussion structure that users assert one's opinion through tweets and provide the rationale or support through url links.

 

Diffusion in Twitter

Further, authors conduct two case studies on exploring information diffusion in Korean Twitter Network. From all the tweets collected for two month period, data for case study were filtered by specific keywords for each case. They picked two cases that are each related to two main function of Twitter; 1) Twitter as an alternative news media which spread breaking news more immediately 2) Twitter as a sort of public sphere for disussion or debate arena where people argue on specific agendas and issues. Reesults of each cases are introduced briefly.

 

Case Study 1 : Breaking News in Twitter - 'heavy rain' in Chuseok Holiday

The best example of Twitter's diffusion of breaking news was related to the heavy rain during the Chuseok holiday on September 21, 2010. Contrary to the weather forecast by the Korea Meteorological Administration, more than 100 millimeters of heavy rain poured in a short period of time, and Twitter's role was by far prominent in line with the situation in which most media were unable to make normal reports due to the holiday. Authors collected total of 41,724 heavy rain-related tweets from 0:00 to 24:00 on September 21 and tracked the spread of news and opinions on Twitter.

 

The number of 'heavyrain'-related Tweets by Hour on 2010.9.21

Above figure shows how many 'heavy rain'-related tweets were written in each hour during the day of September 21, 2010. From midnight to noon, there was almost no tweet, but it began to surge over noon, especially from 1 p.m. to 4 p.m., when heavy rain damage was severe. Especially in the 4pm to 5pm period, the most tweets - 7,611- were written in an hour which is 18.24% of all tweets collected.

After analyzing the pattern of contents that were spread throught the day, one could see that the situation progresses in the order of (1) breaking news (2) cautions (3) policy evaluation on the failure of official alert system (4) encouraging mutual assistance among citizens who got financial damage due to the rainfall. Authors inferred that when everyone had difficulties together, such as natural disaster like sudden heavy rainfall, discussions through Twitter were going on in a fairly reasonable and rational way.

 

Case Study 2 : Opinion Dissemination in Twitter - "ideological consumption" debate

Second case of diffusion was the ‘ideological consumption debate’ regarding the sale of pizzas at Emart, a large supermarket chain. After Emart started sell pizzas in supermarket, some criticized that it would ruin small business pizza places in town. Some twitter users left criticizing mentions on twitter of Chung yong jin, then vice-chairman of Sinshegae - a business group that owns Emart; then Chung replied to some of them saying "That(whether one will buy pizza from Emart or small business) is up to Consumer's choice; Do you consume 'realistically' or 'ideologically'?". This provoked heated discussion in Twitter.

Analyzing the contents of the so-called "ideological consumption" debate confirmed that people were using two methods when the debate took place on Twitter. One method is to argue using opposite concepts or examples within the same framework of argument, and the other is to try changing the framework of the argument itself by introducing a new category that is not used by the opponent.

Moreover, case study on ideological consumption debate reveals relationship between each 'group' of users. Below images are summarized interaction network between each group of users who were each 'critical', 'neutral' and 'advocates(pro)' toward Emart and Chung. Critics and neutralists had a mechanism of mutual reinforcement within their own group and a dynamic in which their opinions are spread to different groups, while advocates have no such social dynamic at all. Specifically, the pattern that advocates follow neturalists but not retweet their tweets implies that relatively minority groups take strategy to react to only those who are clearly against them (critical) or with them(pro) to maintain their ground more easily.

 

Bloc Network Model of Twitter Networks based on each group of attitudes. Follow network(left) and RT network(right)

 

* Orignal Abstract of the paper is attached below.


Abstract

This paper attempts to delineate the structure and dynamics of Korean Twitter network. After identifying 1,133,365 Twitter accounts who were presumably Koreans, we crawled information including tweets, bios, followings, replies, and retweets through Twitter API for the period 1 August 2010 through 30 September 2010, which was then put into social network analysis.

This paper is largely divided in two parts. The first half reports general characteristics of Korean Twitter network, together with characteristics of most-retweeted tweets and use of url links. The second half contains two case studies which became huge social issues during the research period: the heavy rain during the Chuseok holiday season and the ‘ideological consumption debate’ regarding the sale of pizzas at Emart, a large supermarket chain.

Despite this is still a heuristic research, we were able to come up with a few theoretical implications. Twitter is basically a space for communication and discourse. Caution should be used when interpreting the widely-known power-law distribution in terms of social science research. Twitter delineates a strong inclination toward becoming an alternative media vis-a-vis offline traditional media. While RT is a tool for spreading opinions and information, the use of RT is different in a disaster situation from that regarding a socio-political issue. When a debate goes on in Twitter, there are two major strategies. One is to use opposite concepts in the same discourse category, and the other is to bring in a new discourse category which has not been used by the contending party.

KeyWords | twitter, network analysis, opinion dissemination, microblog, SNS

댓글