Influencing policy (training slides from Fast Track Impact)
Identifying Influential Twitter Users in Sejong City, South Korea
1. Chien-leng Hsu (Post-doctoral research fellow) Se Jung Park (PhD student) Han Woo Park (Associate Professor) Department of Media & Communication, WCU Webometrics Institute, Yeungnam University hanpark@ynu.ac.kr http://www.hanpark.net http://english-webometrics.yu.ac.kr Presented at the 5th Complexity Conference, 27 Nov 2010, Seoul, Korea Identifying influential Twitter usersThe case of Sejong City in South Korea
2. About this study This research explores influential Twitter users and Tweets by: Using Sejong City (세종시) as the case study Using different measurement methods Studying topics (keywords) mentioned in the Tweets
3. Key Users on Twitter Communities The 2-step flow of communication theory (Katz & Lazarsfeld, 1964) (Online) opinion leaders: determinants of rapid & sustained behaviorchange of members in a community (Valent & Davis, 1999) Park, Jeong & Han (2008) practices (political) participation deligently aggressively expresses opinions
4. Identifying Key Twitter Users The “network structure” approach: fails to identify influential Twitter users (Leavitt et al, 2009; Haddadi et al, 2010) The “actor relation” approach: content reachability Fisher & Gilbert (2009) ➭ replies, retweets, mentions & attributions The majority of users: silent & passive a user’s influence: information forwarding activity (Romero et al, in submission) Trace influential user over time
5. The Sejong City Project The original plan (Moo-Hyun Roh in 2005): To allocate 2/3 of government offices to Sejong, Chungnam (충남) Necessary for regional development The excessive centralization of Seoul & its vicinity➭ limited innovation potential (Shapiro, So & Park, 2010) The revised plan (Myung-Bak Lee): A center for education, scientific research & high-tech industries Partitioning the capital would weaken Korea’s competitiveness & innovation capability
6. Research Questions Who are the influential users who produce Tweets related to the Sejong City project? What are activities of the influential users? What is the relationship between the influential users? What are the keywords frequently used by the influential users in the Sejong City issue network?
7. Data collection & analytical techniques Data collection Dates of collection: 15 March ~ 12 April 2010 Twitter scraper: An automated computer program to retrieve Tweets from Twtkr (twitterkr.com) Twitter API: Twitter user’s public data Analytical techniques Basic data: Location Number of Tweets Lists of followings Lists of followers Pearson correlation test Four posting activities: Normal tweets Being retweeted by others Being replied by others Being mentioned by others Krkwic (keywords analysis)
12. Amendment of Sejong City law & politicians Critical reviews on Sejong City law Controversies & solutions Agreement & social welfare Other social & political issues Conflicts between political parties Political ideologies & concerns on national debt National policies
13. Discussions (I) Influential users include media outlets & ordinary users Correlation tests: The occurrence of Tweets vs. the number of Tweets ➭ significantly correlated (Pearson correlation=0.663, p<.01)➭ Influential users tended to address public issues The number of followers vs. the number of followings➭ significantly correlated (Pearson correlation=0.871, p<.01)➭ Influential users had mutual ties in the network Influential users are likely to act as news brokers & deliver their views in a single-issue community Having mutual relations with other influential users may allow an influetial user to make his/her own opinions available to a wider audience
14. Discussions (II) Referral activities/relationships Media outlets ➭ normal tweets ➭ messages were not circulated well among other users Ordinary users normal tweets, retweets, mentions & replies More likely to interact with the indirect presence of media outlets Keyword network Politicians, government projects & social-political issues mentioned Influential user some keywords specific to his/her cluster similar keywords used ➭ a sense of community