CRM114, named for a Cold War radio receiver used in the movie “Dr. Strangelove,” is a text categorization program that focuses on phrases instead of simply words. It’s used most often to filter e-mail for spam. Unlike other filters that check words in an e-mail against a dictionary of terms that occur frequently in unwanted messages, CRM114 checks against a phrase book. That makes it more nuanced than traditional one-word-at-a-time filters, says Bill Yerazunis, CRM114’s creator. The filter also incorporates a number of other spam-catching techniques, such as Lightstone’s Winnow algorithm, a specialized correlation technique, and hyperspace-distance and bit-entropy classifiers. In addition, it features its own programming language for plugging it into an e-mail architecture.
The antispam man >
