Researchers from the University of California-Santa Barbara have created a free spam-filtering service for Twitter that detects malicious accounts and forwards the information to the microblogging service for enforcement.
The service, dubbed Spamdetector, is the brainchild of Ph.D. student Gianluca Stringhini.
“We started studying social network spam one year ago,” Stringhini said. “With the rise of social network sites, this problem became really important. During our studies, we found out that spam bots (which are likely infected computers) have some peculiar behaviors that [differ] from the [ones] legitimate users have.”
For example, spammers are usually very aggressive when it comes to following users in the hopes they can build up a large pool of followers of their own. This point was underscored by a recent analysis by Barracuda Networks, which found 26 percent of users have at least 10 followers and 40 percent are following at least 10 people.
Spammers also send similar kinds of tweets, Stringhini said. But the service is also helped by users submitting information about suspicious accounts.
“The usefulness of users flagging spammers is that leveraging this information we can ‘target’ the crawling to those profiles that send tweets similar to the ones that have already been detected as spam,” he said. “In this way, we are able to detect more spammers in a shorter period of time.”
Twitter security has taken more than a few hits in the past two years as the number of users has soared. In response to the growing abuse of URL shortening services to send malicious links, Twitter recently launched a URL scanning service.
“It is not easy to reliably detect malicious users, because there’s always the risk [of suspending] legitimate accounts by mistake,” the researcher said. “However, the Twitter spam detection improved a lot during the last months. Many spam profiles get deleted even before we flag them to Twitter.”
To sign up for the service, follower the user @spamdetector on Twitter.