All that Twitters is not gold

Researchers in HP Labs have recently published thought-provoking studies on important trends related to Twitter, one of the most widely used social-media tools today.

Rise of the Twitter cyborg

Miranda Mowbray, a senior researcher in Automated Infrastructure Lab in Bristol, England, presented her paper, “The Twittering Machine,” at the Computer, Freedom and Privacy Conference this summer in San Jose.

Mowbray observes that Twitter is meant to be a medium that people can use to express themselves, but automated machines threaten to flood the network with unwanted tweets.

Her paper offers recommendations for curtailing Twitter spam.

Read “The Twittering Machine.”

Tweets and retweets

Also this summer, Bernardo Huberman, director of the Social Computing Lab in Palo Alto, collaborated with a member of his team, Sitaram Asur, and others to publish “Influence and Passivity in Social Media.”

Based on an analysis of 22 million tweets, and using a special algorithm, the team discovered that having a large number of followers on Twitter (being “popular”), does not necessarily mean that one is influential.

Rather, the most influential users somehow persuade their followers to re-tweet; that is, to forward information to others, thereby propagating it across the network.

Read the paper.

Watch a brief video of Huberman describing the study.

Telling the future with tweets

Can social-media content be used to forecast the box-office revenue of movies, and other real-world future outcomes?

Yes, say Bernardo Huberman and Sitaram Asur of the Social Computing Lab.

In their paper, “Predicting the Future with Social Media,” Huberman and Asur describe how a simple model, which is built on how fast tweets are created, can outperform other market indicators.

The authors convey that social media “expresses a collective wisdom which, when properly tapped, can yield an extremely powerful and accurate indicator of future outcomes.”

Their method has implications not only for movies, but for product marketing, election outcomes, and more.

Read the study. 

Explore other papers by researchers in the Social Computing Lab.