Spam-Fighting Software Learns To Tell Good E-Mails From Bad Claims 99-Percent Junk Mail Reduction Eric Grevstad
Mon 7/14/03 -- Audiotrieve LLC says its founders, former executives from Dragon Systems, have applied the natural-language technology used by speech-recognition software to the problem of junk e-mail. The result: InBoxer, a new type of spam-fighting software that claims to reduce unwanted e-mail by 99 percent.
Working within Microsoft Outlook 2000 or 2002/XP (not Outlook Express), InBoxer doesn't use conventional filters -- although users can define trusted senders or companies -- but learns by example, using the messages in Outlook mail folders to build a personal language model. The company claims InBoxer can learn to tell the difference between an order acknowledgement and sales pitch from the same company, or between a romantic note from your spouse and racy spam from a bulk mailer.
Messages are quarantined, not deleted, so you can teach InBoxer to allow newsletters or other messages you like, even if other filters might consider them spam. The program is priced at $30, with a $25 introductory offer through October 1.
