Who says what to whom on twitter

Xiaomei Wu

Winter A.Mason

Jake M.Hofman

Duncan J.Watts

Main research of this paper

Introduce a method for classifying users using Twitter Lists into “elite” and “ordinary” users, further classifying elite users into one of four categories of interest—media,celebrities, organizations, and bloggers.

Investigate the flow of information among these categories, finding that although audience attention is highly concentrated on a minority of elite users, much of the information they produce reaches the masses indirectly via a large population of intermediaries.

Find that different categories of users emphasize different types of content, and that different content types exhibit dramatically different characteristic lifespans,ranging from less than a day to months.

Motivation of this paper

Theories of communications have tended to focus either on “mass” communication or on “interpersonal” communication

New channels of Mass communication :cable television, satellite radio, specialist book and magazine publishers, sponsored blogs, online communities,and social news sites.

New channels of interpersonal communication :personal blogs, email lists, and social networking sites

Masspersonal channel:Twitter

Related work

Kwak et studied the topological features of the Twitter follower graph, number of followers, page-rank, and number of retweets

Cha et compared three measures of influence—number of followers, number of retweets, and number of mentions

Weng et compared number of followers and page rank with a modified page-rank measure which accounted for topic

Bakshy et studied the distribution of retweet cascades on Twitter

Innovation of this paper

Shifting attention to the flow of information among different categories of users. Focus on 4 identifying specific categories of “elite” users:media,celebrities, organizations, and bloggers.

Data and methods

Data set

Twitter follower graph

Twitter firehouse Twitter list

Twitter follower graph

Observed by Kwark et in July 31st,2009 Included 42M users and 1.5B edges Follower graph network is a directed network

characterized by skewed distributions both of in-degree(followers) and out-degree(friends)

Twitter firehouse

5B tweets generated over a 223 days from July 28,2009 to March 8,2010 from the Twitter

Focus on the subset of 260M containing bit.ly URLs

Twitter lists

Conclusions

Who listens to whom Who listens to what Two-step flow of information Lifespan of content Lifespan by category

Who listens to whom

0.05% of the population accounts for almost half of all posted URLs.

Attention is highly homophilous-celebrities following celebrities, media following media, and bloggers following bloggers.

Who listens to what

Category:World News,U.S News,business, sports,Heath,Technology,Science,Arts

organizations show little interest in business and arts-related stories, and high interest in science, technology, and possibly world news. Celebrities, by contrast, show greater interest in sports and less interest in health, while the media shows somewhat greater interest in U.S. news stories.

Two-step flow of information

Half the information that originates from the media passes to the masses indirectly via a diffuse intermediate layer of opinion leaders

Lifespan of content

Different types of content exhibit different lifespans Classic music videos,movie clips,long-format magazi

ne articles have long lifespan than daily news stories

Lifespan by category

For vast majority of URLs,longevity is determined by rediscoving

For URLs introduced by elite users,longevity is determined by retweet

Strength & Weakness

Use twitter as their research object

Classify twitter lists into elite users and ordinary users

Emphasize elite users

Restrict attention just to URLs on Twitter

Overlook the unanticipated categories that may be of equal or greater relevance than the selected four categories

Future work

Apply similar methods to quantifying information flow via more traditional channels, such as TV and radio

Explore automatic classification schemes from which additional user categories could emerge.

To extract content information in a more systematic manner—the “what” of Lasswell’s maxim; and second, to focus more on the effects of communication by merging the data regarding information flow on Twitter with other sources of outcome data.

Thank you