Researchers at the University of California, Riverside have created a Facebook application called MyPageKeeper to stop hackers from hijacking pages and posting spammy links. In a four-month study, they found that the app caught 97 percent of the social malware that turned up during the experiment.
From June to October 2011, the team of engineering professors and graduate students sorted through 40 million posts from 12,000 people who had installed MyPageKeeper to see how well the app was working. During that time, 49 percent of users had been exposed to social malware at least once.
“Malware on Facebook seems to be hosted and enabled by Facebook itself,” Faloutsos added. “It’s a classic parasitic kind of behavior. It is fascinating and sad at the same time.”
The researchers are now using the word “socware,” pronounced “sock-wear,” to describe parasitic or criminal behavior that’s specific to social media sites. Twenty percent of the links are hosted inside of Facebook.
Socware is slightly different from email spam. For one thing, the posts mimic status updates from friends rather than official notices from companies. The acronym “omg” is 332 times more likely to appear in Facebook socware than in email spam, the study found. Similarly, the word “bank” is 56 times more likely to turn up in e-mail spam than it is on Facebook.
During the study, the researchers were also surprised to find that only 54 percent of scammers used a link shortener like bit.ly to hide their obviously spammy links, which had URL names that promised free iPhones or NFL jerseys. People must not have looked too closely at the links before they clicked on them.
On average, MyPageKeeper takes about .0046 seconds to go though and mark the posts as safe or unsafe and alert the users. It looks for keywords that are indicative of spam, such as “FREE,” “hurry,” “deal,” and “shocked.”
The app also uses social context clues, such as the number of likes and comments a post has, to weed out the spam. Usually, people do not post comments or “like” social malware posts, the study found.
In the end, the app incorrectly identified socware only 0.005 percent of the time.
Image by Amy Walters via Shutterstock.