Indiana University

Skip to:

  1. Search
  2. Breadcrumb Navigation
  3. Content
  4. Browse by Topic
  5. Services & Resources
  6. Additional Resources
  7. Multimedia News

Last modified: Thursday, July 28, 2011

NSF awards Truthy team $905,000 to develop tools for real-time social media analysis

July 28, 2011

BLOOMINGTON, Ind. -- Indiana University's Truthy team -- four informatics and computing professors who last year received international media attention after establishing a watchdog website to identify political astroturfing occurring via Twitter -- has received a National Science Foundation grant of $905,000 to broaden work analyzing the massive stream of public data found in large-scale social media networks.

Network of Hashtags

This single-frame capture of a diffusion network was taken from an animation of tweets analyzed during a period of the Egypt uprising. The Truthy website now has a movie tool which allows people to create their own animations of diffusion networks for memes or events of interest.

"To better understand how and why information spreads online, we intend to develop a framework that allows us to apply the same analytical methods to a broad variety of data feeds, including Twitter, Google Buzz, Google , Yahoo! Meme, and Facebook," said Filippo Menczer, who joins Alessandro Flammini, Alessandro Vespignani and Johan L. Bollen as IU Bloomington School of Informatics and Computing professors serving as principal investigators on the grant. "We expect that this new computational framework will offer an unprecedented level of data interoperability for the real-time analysis of a social media data stream on the order of millions of posts each day."

Even with advances in the field of information diffusion in recent years, IU researchers said there remained a need for commonly accepted, detailed and empirically validated models that can reliably predict the size and scope of specific diffusion processes. One goal of the research will be to create a model that captures the shared traits of information diffusion processes in different social networking sites while still accounting for their diverse structures and interfaces.

"We can achieve this by modeling a stream of social data as a series of time-stamped events that represent interactions between actors and memes," Flammini said. "The hope is that the new framework will help social networking researchers move away from a series of compelling anecdotes and build a true science of online social interaction."

The researchers would like to determine whether there are different, definable types of information-spreading behavior, and if so, whether they might be identifiable through formal mathematical models.

"One way we hope to advance those models would be to improve sentiment-tracking tools that too often rely on classifications in either positive or negative categories and that ignore the complex nature of human mood states," Bollen said. "Instead, a set of novel sentiment- and mood-tracking methods focused on empirical cross-validation against other indicators -- stock market, weather and news event data -- could be used for meme characterization."

Then researchers might be able to determine what factors were in play when some memes go viral, while others did not propagate. Those factors could include the community structure of online users, the network topology, occurrence with other memes, or the appearance of certain semantic or affective features.

"There are a number of websites that have emerged recently that track popular and trending memes, but they are not designed for scientific purposes," Vespignani said. "We hope to provide for the first time a large-scale infrastructure to delve deeply into a broad set of questions about how and why information spreads online."

The researchers will eventually create and maintain a web service that allows people to follow trends, bursts and suspicious memes, much as they did with the politically-oriented last year, and which they say could have the potential for detecting hate speech or subversive propaganda and for preserving open debate.

Visitors can visit to browse animations of meme diffusion networks -- for example, the researchers have created one connected to the uprising in Egypt -- and they can also enter any hashtag being tracked by Truthy, a specific time period, and a network type and the system will generate a meme diffusion movie and upload it to YouTube. To view the Egyptian uprising twitter meme diffusion animation, click here.

For more information or to speak with any of the researchers, please contact Steve Chaplin, University Communications, at 812-856-1896 or