Miles Keaton <[EMAIL PROTECTED]> writes: > Any benefit to having two different kinds of spam filters, so that one > (SpamAssassin?) goes system-wide, and another (Dspam?) is the > user-trained one?
SpamAssassin plus user-training works better than dspam. Plus, you only have to understand and integrate one thing, not two. I don't really see much benefit to installing two systems. If some users want to use SpamAssassin only trained on their own data, then you can just rely on the BAYES_NN rule hits. SpamAssassin's Bayes component is itself a great Bayesian filter. Reference: http://plg.uwaterloo.ca/~gvcormac/spamcormack.html If you don't want to use SpamAssassin for some reason and want to go pure Bayes, I'd suggest bogofilter, spambayes, or spamprobe (although I have no personal experience with spamprobe). Of course, I think SpamAssassin is better. :-) Daniel -- Daniel Quinlan http://www.pathname.com/~quinlan/
