Pattern recognition techniques allow products to learn from past examples, helping them to formulate a way of classifying messages as "spam" or "legitimate." The security company plans to use the technology as one measure of how likely it is that a particular e-mail message is spam.
"We already have five levels of detection," said Tim Smithson, solution marketing manager for Network Associates. "This is just adding another layer of detection...based on the spam it is currently seeing."
Network Associates is the latest company to use Bayesian techniques to enhance its product's ability to recognize patterns. Microsoft intends to use the technology to build a better search engine. America Online built Bayesian techniques into its antispam filter for the latest version of its software.
Based on the theories of ex-clergyman Thomas Bayes, Bayesian filtering essentially attempts to create a spam filter based on the characteristics of legitimate and unsolicited bulk e-mail received in the past.
The appeal of the Bayesian technique is its deceptive simplicity. The predictions are based completely on data culled from reality--the more data obtained, the better it works. Another advantage is that Bayesian models are self-correcting, meaning that when data changes, so do the results.
The technology has already been built into Spam Assassin, an open-source spam filter created by security firm Deersoft, which Network Associates bought in January 2003. Network Associates tests out changes to the software in the open-source community before introducing it into its own products, Smithson said.
"What we like to do with open-source code is let it get tested in the community for a while, and then we add some new features when we put it in our product," he said.
The Bayesian filtering has been added to the most recent version of SpamKiller, and clients who subscribe to the company's updates can get it automatically, Smithson said.
Play video
There are currently no comments for this post.