JM>> Why does the Bayesit plugin not work very well? It will pick up JM>> an occasional spam but misses most. It will still fall for the JM>> same messages even after I mark them as spam. ????
M3> Spammers have gotten "hip" to how Bayesian filtering works and have M3> devised ways to get around it by filling the message with random words M3> and the like, so that no two messages are the same ... just as they M3> were able to get around "name" filtering by creating fictitious M3> addresses that were always different. M3> It is a constant struggle. We find a new "tool", then "they" find a M3> way around it <sigh> M3> Bayesian filtering still works, just not as effectively against the M3> rabbid spammer who only seems to care about getting a nonsense message M3> into your inbox ... as opposed to the commercial spammer who does want M3> you to buy something (what are they thinking? <shrug> ) That's a factor, but I found BayesIt just didn't work for me. It seems to work for some here, though. K9 is a Bayesian filter, too, and it's operating at 98.6% accuracy for me. BayesIt was less than 1% accurate after training for more than a month with hundreds of good and spam messages. -- Code 2 :canadaflag: mailto:[EMAIL PROTECTED] The Bat! version 2.12.00 on Windows XP Service Pack 2 On a New York convalescent home: For the sick and tired of the Episcopal Church ________________________________________________ Current version is 3.0.1.33 | 'Using TBUDL' information: http://www.silverstones.com/thebat/TBUDLInfo.html
