Grad student goes undercover to infiltrate “Water Army”

A virtual flood is on the rise. There’s a growing “water-army” of online posters at work who earn their name by flooding websites with fake comments and paid endorsements in an orchestrated deluge. And now a UVic graduate student wants to ensure that you don’t take a bath from their cyber confidence tricks.

Cheng Chen is a second-year master’s student in the Department of Computer Science. With his co-supervisors and co-authors, Dr. Kui Wu and Dr. Venkatesh Srinivasan, Chen is developing a systematic approach to detect these hidden paid posters.

And he went underground to do it.

Posing as one of the legion of unemployed students or underemployed workers in China who sign on with  PR firms as paid posters, Chen got a job. The qualifications were few, according to Chen.

“All you need is a credit card in order to receive payment,” says Chen. “It is hard to accomplish the job because you need to have many different account IDs—that’s the time consuming element.”

There’s no shortage of work as a paid poster, says Chen.

“There are many, many tasks, and you can choose any of them. Every task is to post something,” says Chen. “Sometimes you post to forums to say good things about products, or you post comments to blogs. [The employer] gives you articles, which you don’t need to modify, to post to specific blogs. After that, they hire other people to post comments to the blogs to make the articles more popular.”

The pay is generally mere pennies per task, says Chen, so it’s not a get-rich-quick scheme.

“Some tasks will give you points and you can use the points to get money,” he says. “Or you can store the points in your account, accumulating points to get more money.”

By getting to know the behavioral practices and organizational structure of paid posters through his undercover research, Chen and his supervisors designed and validated a new detection mechanism.

They tested it on comments and reports on two news websites, Sina.com and Sohu.com, concerning a conflict between two competing Chinese tech companies, Tencent and Qihoo 360.

“We wanted to find patterns of behavior of the online users, because we know there are two groups of people—normal users and paid posters. In our paper, we describe some different behaviors between these two groups of people,” says Chen.

The team found differences in the kinds of comments posted by members of the water army and regular users, and differences in the percentages of replies, the time of posts, the duration of activity and the number of reports commented on by the posters.

According to the researchers, the trick is to stay ahead of the water army by developing new systems that prevent paid posters from learning how to avoid detection.

“It’s like a game of cat and mouse,” says Wu.

And it’s a high-stakes game involving millions of players.

Proliferating online along with the water army are shill bidders on eBay, spam reviewers on travel sites, floggers (fake bloggers), and virtual astroturfers—so-called for their attempts to generate fake grassroots support for causes and companies.

While the researchers have been unable to confirm the veracity of specific online comments and posts, they want to help consumers and citizens kept afloat despite the sea of malicious marketing.

“The idea is to ultimately have a tool that can help people identify paid posters, because it is very hard as a user to guess, when reading a comment, if it came from a genuine user or from a paid poster. The goal of our research is to build a system that can identify who are potential paid posters and possibly create an online service that can help people,” says Srinivasan.
Next up: the research team is taking on popular FAQ sites, like Yahoo! Answers, to detect coordinated groups of paid posters working together.

In the meantime, for those who are a little wet behind the ears: buyer beware.

Read the team’s report, “Battling the Internet Water Army: Detection of Hidden Paid Posters,” at: http://arxiv.org/abs/1111.4297.