Building a better spam trap

Unsolicited e-mail messages, or spam, are on track to make up the majority of traffic on the Internet. But a group of researchers and developers gathered here Friday hopes to halt that by coming up with better ways of blocking those messages from consumers' in-boxes. The Spam Conference, held at the Massachusetts Institute of Technology, was originally intended to be an informal gathering of 30 people or so. But more than 500 registered to discuss and debate the best way to battle the problem.

"Spam-filtering is shooting at a target that is not just moving, it's taking evasive action," said Bill Yerazunis, a research scientist at the Mitsubishi Electronics Research Lab and the author of the CRM114 Discriminator, a spam filter.

Earlier methods of fighting spam focused on certain aspects: blocking e-mail from an address, or with the word sex in the subject line, for example. But spammers have found many ways to bypass these methods. More recently, spam-fighters have been looking at statistical analysis to help determine whether a message is spam or not. Programmers can use large archives of known junk e-mail to search for patterns and properties, then use those results to test incoming mail.

Those filtering methods can help combat one problem in spam-fighting: false positives, which lead legitimate e-mail to get mislabeled as spam. And the methods may also be more difficult for spammers to fight than traditional defenses.

View: The full story