340 likes | 605 Views
Social Graph and Analysis of Social Network. YOSHIDA, Masami Chiba University – Chiba, Japan Graduate School of Social Sciences and Humanities, Studies of Public Affairs, Fields of Public Education. Related links and sources of this presentation are appeared in Twitter.
E N D
Social Graph and Analysis of Social Network YOSHIDA, Masami Chiba University – Chiba, Japan Graduate School of Social Sciences and Humanities, Studies of Public Affairs, Fields of Public Education Related links and sources of this presentation are appeared in Twitter. Also, your comments and questions are acceptable on live with a hashtag of ……………..
Educational Orientation • BehavioristOrientationTeacher OrientedPreparation: Objectives, Text, Materials, EvaluationEDDIE, ASSURE • Constructivist OrientationStudent OrientedOutcome OrientedProblem Solving by Discussion5E, 7E • Social PsychologySocial Graph
Introduction • Globalization and Flat World in economical transactions, information communication and cultural exchanges. (Freidman, 2005) • Changes in how we communicate and interact with information • The social relationships have been altered by online communication in terms of scale and size.
Purpose of the Study • To monitor Japanese subcultural international project. • To know how public opinion is formed through inter-cultural understanding by prolonged monitoring of SNS. • To imagine public opinion in the future internationalization of Japanese culture. • To clarify how a new mode of communication emerged. As Method of Analysis • Social Graph
Target Community • The JKT48, an idol group that is active in Indonesian music business by applying a music business format of Japan. • In Sep. 2011, Japanese fans emerged in SNS, and started to exchange online messages via a BBS; 2CH, the largest BBS in the world. • 2CH: Thread floating bulletin boards, 230 million page views/day, 2 million posts/day • Record to monitor fans’ communication for 1.5 year is used.
Data and Analysis • BBS record =>dat file =>convert to xml file by VBA scripts => XML file => Microsoft Excel • Analyzed by NodeXL; an extendible toolkit for community exploration • Visualizing social graph and calculated graph metric.
‘Graph’ as Deductive Way of Investigation The social graph draws the personal relations of online users as a style of graph. This graphin this context is made up of ‘vertices’ or ‘nodes’ and ‘lines’ called ‘edges’ or ‘ties’ that connect vertices.
Network Structure In-Edge Out-Edge Conceptual structure • Vertex(Node): Usually human • Edge(Link): Usually message, and can be labeledas directed/undirected. Characteristics • Hubs: Important vertices • Bridge: An edge that connects two separate groups • Cluster: A group of vertices Vertex Directed Network
Complex Networks • This is the academic field to study the graph with non-trivial topological features. • There are three known characteristics in complex networks researches. • Scale Free Network • Small WorldPhenomenon • Clusters • There is no effective mathematical model to view above three characteristics
a. Scale Free Network Random network is Gaussian distribution and the most scientific studies in education followed this basisso far. But, Barabashi found social activities of human followed Power Law. This includes existence of hubs. • Barabasi & Albert (1999), Emergence of Scaling in Random Networks, Proc. of Science, 286(5439), pp.509-512
Longtail Gaussian Distribution Power Law ranking Power Law Gaussian Distribution ranking
b. Small World Phenomenon The people who may be very far apart physically and socially are still connected with relatively small paths. Many vertices in social community but small distance one another. Milgram: “Six degrees of separation” • Milgram, S. (1967), The Small World Problem, Psychology Today, 2(1), 60-67. • See Bacon number at http://goo.gl/w0azj • Kleinberg (2000), The Small-World Phenomenon: An Algorithmic Perspective, Proc. of the 32th ACM Symposium on Theory of Computing, 163-170.
Weak Ties • “The Strength of Weak Ties”: A highly influential sociology paper, with over 27,000 citations.http://goo.gl/RJZZ5i • Granovetter surveyed people in a suburban community about their channels of mobility information. • Often: at least once a week, 16.7% • Occasionally: more than once a year but less than twice a week, 55.6% • Rarely: once a year or less, 27.8% • Weak ties play a role in affecting social cohesion.
c. Cluster In scale free networks by a process of "preferential attachment", in which new network members prefer to make a connection to the more popular existing members. This is called as Watts=Strogatz model or Small World Network Model. • Growing divide • Watts & Strogatz (1998), Collective Dynamics of Small-World Network, Proc. of Nature, 393(6684), 440-442.
Target Data The authors selected three separated months and analyzed changes of communication. • September, 2011: At the dawn of the online community, just before selection of JKT48 members by audition. Communication in BBS was processed under the lack of information. • February, 2012: JKT48 had appeared frequently on TV, news and Internet. • January, 2013: They had a theater in a shopping mall that dedicated to JKT48, and offered daily show by two teams. Note: Author did not post any message on the target threads of 2CH.
Bridge Influencers • Extracted bridge influencers that posted information of another community • Two types of bridge influencers would exist potentially • A user from Indonesianfans’ community • A well-informed Japanese fans who live in Indonesia
Changes of Community Size During 17 months, vertices increased 13 times and unique edges increased 21 times where messages increased 8 times. This means that spectators had been increased. On the other side, diameter was increased and it reached to 5 times. This implies existence of clusters in a later graph. Growth of community: Communication messages >Posted messages
Social Graph, Sep.2011 • Viewed by Harel-Korenfast multiplex algorithm • Vertex size proportional to in-degree statistics • Two groups are confirmed • Vertices color of bridge influencers mapped red
Bridge Influencer, Sep.2011 • Bridge influencers are pulled out to the peripheral area manually • Viewed by Fruchterman-Reingoldaalgorithm • Bridge influencers located peripheral or contiguous to a hub.
Social Graph, Feb. 2012 • Viewed by Harel-Korenfast multiplex algorithm • Few groups are confirmed. • Showed characteristics of scale-free network. • Few clusters are appeared
Bridge Influencer, Feb. 2012 • Viewed by Fruchterman-Reingoldalgorithm • An Indonesian fan of bridge influencer was a hub, but not in a largest cluster. • Japanese fans of bridge influencers still located in peripheral.
Social Graph, Jan. 2013 • Viewed by Harel-KorenFast multiplex algorithm • An Indonesian bridge influencer is not a hub of the largest cluster. • Average posting rate 5.01 posts/vertex in Sep.2011 was decreased to 3.09 posts/vertex in Jan.2013. • But, edge rate was increased 0.23 edge/message in Sep.2011 to 0.65 edge/message in Jan.2013. • Utterance => Message
Bridge Influencer, Jan. 2013 • Viewed by Fruchterman-Reingold algorithm • Japanese bridge influencers still located in peripheral.
Comparison of Bridge Influencers • Indonesian bridge: posted popular topic of conversation • Japanese bridge: posted information everyone did not know, stay in peripheral strategically • Remarkable characteristics are seen in the scores of eigenvector centralitythat shows Indonesian fans locate close to a hub.
Relation with Other Community • In Jan.2013, Indonesian SNS sites with rich information were noticed by users. • Online users gradually began to collect information by themselves. • Presence of bridge influencers were decreased
This study was partially supported by the research project. Grant-in-Aid for scientific research (B) of JSPS, project number 26301035. YOSHIDA Masami Professor of Graduate School of Humanities and Social Sciences, Chiba University, Japan yoshida-m@faculty.chiba-u.jp Advisor of Thailand Cyber University Now on Sabbatical Leave Stay in Silpakorn University