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: Monday, October 25, 2010

While data mining for political astroturfers, truthy.indiana.edu is hitting pay dirt

FOR IMMEDIATE RELEASE
Oct. 25, 2010

BLOOMINGTON, Ind. -- Truthy.indiana.edu, the website created by researchers at Indiana University Bloomington's School of Informatics and Computing to root out Twitter-based political astroturfing campaigns, is finding success.

In what the researchers are calling one of the most egregious examples, the Twitter account @PeaceKaren_25, while concealing its ownership identity, generated more than 10,000 tweets since the end of June. Another account, @HopeMarie_25, that was created 10 minutes later and that also has a concealed owner identity, retweets all tweets generated by @PeaceKaren_25 while producing no original tweets of its own.

PeaceKaren

A visualization of the meme diffusion network where @PeaceKaren_25 is connected to @HopeMarie_25 (both represented by black nodes) through the very thick blue (retweet) edge, indicating that @HopeMarie_25 retweets @PeaceKaren_25 constantly.

"The names and behaviors of the two accounts suggest that they are colluding and are most likely controlled by the same entity," said IU Bloomington School of Informatics and Computing Associate Professor Filippo Menczer. "We do not know if these are bots (software that carries out automated tasks) or human-bot hybrids."

Information scientists at IU are interested in understanding the diffusion of all types of memes (ideas or patterns passed by imitation) and are currently using Truthy.indiana.edu as a research tool that combines data mining, social network analysis and crowdsourcing to review political memes and possibly uncover deceptive tactics and misinformation leading up to the Nov. 2 elections.

Menczer said almost all of the more than 20,000 tweets from @HopeMarie_25 and @PeaceKaren_25 support Republican candidates, especially U.S. House GOP leader John Boehner, whose account @GOPLeader is "very frequently retweeted or mentioned."

The research found that tweets by @PeaceKaren_25 and @HopeMarie_25 frequently include links to various websites supporting GOP candidates, and also to Boehner's website gopleader.gov, his Facebook page and blogs, and to the gop.gov website for Republicans in Congress. "Both accounts promote the same targets while the second account also promotes the first account," Menczer said. "This is very clever and hard to catch automatically because it 'looks' real."

The Truthy team has identified several other suspicious memes, displayed in a gallery on the site (http://truthy.indiana.edu/gallery). A burst of activity surrounding the "#ampat" hashtag, popular among conservative users, was driven by two accounts controlled by a user from Illinois.

"Steven Tucker generated over 41,000 tweets this way," said Jacob Ratkiewicz, a graduate student who is developing the Truthy detection system.

Another meme spotlighted by the Truthy project is a website that displays pro-Sarah Palin and anti-Muslim propaganda. Michael Conover, a graduate student who is analyzing the Truthy data, noted that the site includes a graphic video of beheadings by the Taliban. "Most of the tweets originate from one account, @GoRogueRunSarah, which has generated over 15,000 tweets," he explained.

Following a tip from a user who flagged a handful of suspicious tweets smearing Chris Coons, the Democratic candidate for U.S. Senate from Delaware, the researchers uncovered a network of about 10 bot accounts. These bots have names like @krossnews, @BethlehemTweets, and @kingdomcast. They inject thousands of memes, all of which link to posts from the Freedomist.com website.

"To avoid detection by Twitter, duplicate tweets are cleverly disguised by adding different hashtags or subtly tweaking the web addresses," said IU research associate Bruno Gonçalves, who is mining the stream of tweets. "This gives the appearance of a lot of different people sharing the same views, while in reality the bots are flooding the Twittersphere with one coherent political message."

Generating traffic is important, explains Gonçalves: "While usually referred to as 'viral,' the way in which information or rumors diffuse in a social network is different from infectious diseases. Rumors gradually acquire more credibility and appeal as we become more exposed to them. After some time, a threshold is crossed and the rumor becomes so widespread that it is considered as 'common-knowledge' within a community and hence, 'True.'"

Indeed, each of the bot accounts has hundreds, and in some cases thousands of followers, who retweet and spread the truthy memes, he added.

Who is behind all this? According to Menczer the answer is easy in this case: "Most of the bot accounts in this network can be traced back to Bill and Paul Collier from Pennsylvania, who also run the Freedomist.com website." But he also believes this is just the tip of the iceberg.

Menczer said it was unfortunate that all examples of the most egregious 'truthy' memes were from the conservative right, but searches are conducted bias-free with respect to political rhetoric.

"We looked really hard for any 'truthy' memes from the left, but unfortunately we could not find any. I am sure this is not because they don't exist, but rather because there is just very little traffic on Twitter from the left this political season," he said. "The conservative side is creating huge amounts of traffic. Almost all of the most popular hashtags, the most active accounts, and the most tweeted URLs, are from the right. So a large majority of what we are able to track is from the right -- both authentic and 'truthy.'"

The appearance of a grassroots campaign where multiple people independently tweet about these candidates should be considered astroturfing, Menczer said, and the fake accounts have also succeeded at creating "Twitter bombs," leading Google searches for candidate names to return these tweets in the first page of results.

The activities would seem to violate Twitter account rules that forbid posting duplicate content over multiple accounts, forbid a person from impersonating others in a manner that does or is intended to mislead others, and forbid serial accounts for disruptive or abusive purposes, or with overlapping use cases. The Twitter Rules site does state that "mass account creation may result in suspension of all related accounts. Please note that any violation of the Twitter Rules is cause for permanent suspension of all accounts."

For more information or to speak with members of the truthy.indiana.edu team, please contact Steve Chaplin, University Communications, at 812-856-1896 or stjchap@indiana.edu.