data they yield about people’s interactions, have emerged as a valuable new lens through which to explore the full range of communication among individuals. Duncan Watts discussed work at the intersection of social science and computer science performed at Yahoo! that used a subset of Twitter user information and updates.1

Media research has tended to focus on two types of communication— individual organizations broadcasting to large, undifferentiated audiences, and individuals communicating with each other—but generally has not looked at anything that happens between these two extremes. Although most people think of Twitter as a social network, it can also be viewed as a full-spectrum media ecosystem.2 Twitter communications cover the spectrum between the two types of communication traditionally examined by media research; individuals as well as traditional mass media outlets are able to broadcast information. New forms of interaction have emerged, such as mass personal communication, in which “elite” individuals—celebrities, politicians, journalists, or recognized experts— not only broadcast information to large audiences but also engage in public conversations that are widely followed.

One of the biggest challenges to using social media, Watts noted, is the large number of accounts and the volume of data they generate. It is difficult to categorize the more than 200 million Twitter accounts as those associated, for example, with individuals or organizations. In 2009, Twitter introduced a new feature called lists, which provided users with a mechanism for filtering incoming feeds and other users, providing researchers with data (which is public by default) on how users classify each other.

Watts explained that the Yahoo! study drew on a collection of data originally used by Haewoon Kwak in his study of Twitter. Collected in 2009, the data included 42 million users and 1.5 billion individual connections. 3 (However, the focus of the work was on 260 million tweets that included a bit.ly URL, a URL-shortening service.)

One important finding from this work was that a small number of “elite” users were followed by half of all Twitter users. Yahoo! used the list feature to help separate out four categories of elite users: celebrities,

__________

1 Shaomei Wu, Jake Hoffman, Winter Mason, and Duncan Watts. Who Says What to Whom on Twitter. 20th Annual World Wide Web Conference, Association for Computing Machinery, Hyderabad, India, 2011. Available at http://research.yahoo.com/pub/3386.

2 Haewoon Kwak, Changhyun Lee, Hosung Park, and Sue Moon. What is Twitter, a social network or a news media? Available at http://product.ubion.co.kr/upload20120220142222731/ccres00056/db/_2250_1/embedded/2010-www-twitter.pdf.

3 Haewoon Kwak originally made the data public at http://an.kaist.ac.kr/traces/WWW2010.html. However, a change in Twitter’s terms of service resulted in the researchers being unable to share their original data set.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement