SpamSieve 1.3 Offers Bayesian Spam Filtering

Michael Tsai has released SpamSieve 1.3, a spam filter that uses a probabilistic approach to identifying spam based on good and bad messages with which you’ve trained it, much like Apple’s Mail. The beauty of SpamSieve is that it works with many different email programs under Mac OS X (including Entourage, Mailsmith, PowerMail, Emailer running in Classic mode, and, for mail that remains in your In box, Eudora 5.2) and lets you deal with spam inside your email program rather than in a separate application. It also works with any number of accounts and filters mail from any source your email program supports. Though it does require that you train it with a batch of good and bad messages, SpamSieve has proven quite accurate in some of our real-world testing, offering as much as 90 percent accuracy and only 0.5 percent false positives. (Most of the false positives were solicited commercial email, and thus likely to run afoul of SpamSieve’s filtering until it has been trained to recognize similar messages. Other TidBITS staff members have seen less reliable results, however.) New in SpamSieve 1.3 are increased resilience to the ways spammers are now obfuscating common words, the capability to use email addresses in Apple’s Address Book as a whitelist, editing of SpamSieve’s corpus of words, type-to-select in the Corpus window, and the capability to see statistics from after any given date. Despite its incomplete and somewhat clunky integration with Eudora (which is not SpamSieve’s fault, and which will be addressed in a future version of Eudora), we’ve still found it useful and well worth a look. SpamSieve 1.3 is $20 shareware (upgrades from previous versions are free) and is a 1.5 MB download.