Twitter needs spam filters to block counterfeit medicine ads

Evidence that suppliers of counterfeit and otherwise illegal medicines were starting to use the microblogging site Twitter to promote their wares first emerged last year, and has now been reinforced by a new study from researchers at Akron University in the USA.

"Spammers have pounced on this new opportunity for a fast and dishonest dollar, disseminating pornography, fake lotteries, fake inheritances, counterfeit watches, free software, and illegal or ineffective pharmaceuticals," write the study authors Chandra Shekar, Kathy Liszka and Chien-Chung Chan.

The changeover to Twitter and other social media platforms such as Facebook follows a major decline in unsolicited emails peddling prescription drug in the first half of 2011, thanks in large part to the takedown last year of two major spam operations - Spamit.com and the Rusteck botnet - which crippled the operations of rogue online pharmacies.

While pharmaceutical spam "has not caught on as rampantly in Twitter" as in email, the threat still exists, they note, citing one anecdotal case in which a search for Viagra picked up one new tweet every 30 seconds.

A key difference between spam on Twitter and spam via email is the pathway of communication, they note. "In the email world … spam comes to the user. In the social networking model, users come to the spam."

Spammers use a number of techniques such as hashtags for unrelated topics in order to bring their tweets - and the links to ecommerce sites - in front of Twitter users. They also set up fake accounts to try to gather followers, and use retweets to get attention to their messages among users who are not following them.

The researchers have developed a classification scheme for pharmaceutical spam tweets based on data mining techniques which examined around 11 million tweets, including half a million from a one-month period between November and December 2010. 

The aim was to try to identify algorithms and decision tree logic that could be used to filter out spam on Twitter in a manner analogous to the spam-filtering capabilities of email clients. They conclude this is possible, but will be more challenging than for email.

"It seems that a reliable tool for spam detection in general, and pharmaceutical spam in particular, is crucial in applying Twitter’s search feature," they note.