Cloudmark, which earlier this year made waves with the launch of a peer-to-peer spam-fighting service, on Monday at Comdex here introduced a predictive Bayesian engine designed to identify and block unwanted e-mail messages by looking at the message structure.
The engine, dubbed Authority, is designed to prevent spam by blocking messages at the gateway before they can enter corporate networks and eat up both computing resources and employee productivity, according to Karl Jacob, CEO of Cloudmark, based in San Francisco.
"Spam is the silent killer of productivity in corporations, sapping time from employees and taking up a lot of computing resources in larger corporations," Jacob said.
Working in conjunction with Cloudmark's SpamNet p-to-p service, which taps a network of 200,000 users to collect a database of spam, Authority predicts whether a message will be spam by looking for mutations in the message structure, Jacob said. Authority leverages Bayesian machine learning and classification techniques to look at the structural elements of e-mail messages. In addition, Authority can improve its effectiveness over time by the information about spam messages it receives from the SpamNet service, according to Cloudmark officials.
"Like DNA, e-mail messages have a certain structure. Authority predicts whether a message is spam, not by looking at words or phrases but by analyzing [messages] at a higher level of structure," he said.
Appeasing corporate needs, Authority does not require a connection to an outside service and is designed to give IT managers control over messages and in how to respond to spam. The system scans messages for spam, then reports a confidence level about messages -- allowing an IT department to tailor the system to its needs, Jacob said.
"Authority allows IT administrators to have detailed control over what happens to messages and when," he said.
The problem of unsolicited e-mail has grown significantly in the past year, outpacing previous years of steady rise, according to Ray Everett-Church, chief privacy officer at ePrivacy Group, a privacy consulting firm, based in Philadelphia.
"This new approach is really a quite interesting and welcome addition to the market because the existing technologies for blocking spam haven't been doing the job," Everett-Church said.
Because spammers are constantly tweaking their messages to fool automatic filters, which typically look to match words or phrases, Cloudmark's higher level structural approach has a great deal of merit, he said.
"By taking a step back and looking at spam in a more fundamental way, looking for not particular words but characteristics inherent to the spam message, you have a different way of pulling apart the message and performing actions on it," Everett-Church said. "[The] approach to deconstruct spam down to some fundamental elements that will always be present is an interesting approach and a promising one."
In a separate anti-spam announcement, e-mail security vendor CipherTrust on Monday announced FirstAct, the company's in-house team of e-mail threat and spam specialists that will provide analysis and policies to fight spam and security threats in real time, according to company officials in Atlanta.
The service is designed to keep CipherTrust customers one step ahead of spam abusers by monitoring global spam activity and studying the tools and techniques used by spammers. FirstAct will provide analysis as well as tools such as filters, signatures, and policies to help protect customers against new spam attacks, company officials said.