TechAndComputer (Apr. 26, 2012) Nearly a year after U.S. Special Forces killed Osama bin Laden, the events of May 1, 2011 remain one of the busiest traffic periods in Twitter history. More than 5,000 tweets were sent per second when Twitter became the first source with news of bin Laden's death. But how did the news break and quickly spread across the Twittersphere?
A team of Georgia Tech researchers, together with colleagues at Microsoft Research Asia and University of California-Davis, looked at more than 600,000 tweets for answers. By analyzing tweets sent during a two-hour time frame beginning just minutes before the first rumor, they found that opinion leaders and celebrities played key roles. Their data also shows that the Twitterverse was overwhelmingly convinced the news of bin Laden's death was true, even before it was confirmed on television.
The study confirms the widely held belief that Keith Urbahn (@keithurbahn), an aide to former Secretary of Defense Donald Rumsfeld, was indeed the first person to break the news on Twitter. His tweet was sent at 10:24 p.m. Eight minutes later, a CBS producer (@jacksonjk) tweeted her own confirmation. When a reporter with The New York Times (@brianstelter) retweeted both reports, the news began to spread more widely.
"Rumors spreading on Twitter is one thing," said Mengdie Hu, a Ph.D. candidate in the School of Interactive Computing who led the study. "Determining if they are true is another, especially in this era of social media and the rush to break news."
To make a determination, Hu and her team used machine-learning methods to examine more than 400,000 English tweets in the sample. If the message mentioned the death as a fact or in very confident terms, it was classified as "certain." If any hesitation or rumors were mentioned, the tweet was sorted as "uncertain." Within minutes of Urbahn's post, 50 percent of tweets were certain. By the time TV networks broke into programming 21 minutes later, nearly 80 percent were already sure that bin Laden was dead. The number peaked to just over 80 percent after TV made it official.
"We believe Twitter was so quick to trust the rumors because of who sent the first few tweets," said Hu. "They came from reputable sources. It's unlikely that a CBS News producer or New York Times reporter would spread rumors of something so important and risk jeopardizing their reputation. Twitter saw their credentials and quickly believed the news was true."
Also, although nearly everyone on Twitter was talking about the news, a group of 100 "elite users" was actually driving the discussion. Nearly 20 percent of all tweets mentioned one of these elite users. Unsurprisingly, media outlets such as CNN, CNN Espanol and the New York Times led the way, especially in the minutes before and after the TV announcement. But within a half hour of the TV reports, celebrities surpassed media mentions and carried the discussion throughout the night. They included comedian Steve Martin and reality stars Kim Kardashian and Paul "DJ Pauly D" DelVecchio of the "Jersey Shore."
"The celebrities weren't the first people to arrive at the party," said John Stasko, Hu's advisor and professor in the School of Interactive Computing. "But they stayed the longest and brought the most guests."
The findings surprised the researchers, especially because the topic was political and the majority of the celebrities had nothing to do with politics.
Hu and Stasko are using the analysis to develop software that can measure moods and influential people on social media. Marketing companies could use the tools while unveiling new products or searching for celebrity endorsers.
Hu will present the findings in Austin, Texas, at the Association for Computing Machinery's (ACM) Special Interest Group on Computer Human Interaction (SIGCHI) conference in May.
Other social bookmarking and sharing tools:
The above story is reprinted from materials provided by Georgia Institute of Technology.
Note: Materials may be edited for content and length. For further information, please contact the source cited above.
Note: If no author is given, the source is cited instead.
Disclaimer: This article is not intended to provide medical advice, diagnosis or treatment. Views expressed here do not necessarily reflect those of TechAndComputer or its staff.