On Thu, 04 Dec 2003 at 21:29 GMT, Andrew A. Raines penned: > Erik Max Francis <[EMAIL PROTECTED]> writes: > >> "Monique Y. Herman" wrote: >> >>> Have you considered running spamassassin or similar to cut down on >>> this noise? > > [...] > >> I've thought about it. I have some misgivings, simply because I'm >> tired of getting any false positives at all. > > FYI, I've *never* seen a false positive using the spambayes > classifier. It makes a nice front-end to TMDA, especially since it's > written in Python. I keep meaning to plug the two together sometime.
The only false positive I've ever seen with spamassassin is for corporate mailing lists that shout, use html, and otherwise misbehave. (foxtheatre.com and hockeygiant.com, iirc). The way my mail is set up, if I'd had those addresses properly identified in tmda, they never would have been directed to spam. > > Pure Bayesian approaches are a little safer IMHO than rule-based > because you don't have to trust someone else's idea of what you > consider ham. > > -Drew > -- monique _____________________________________________ tmda-users mailing list ([EMAIL PROTECTED]) http://tmda.net/lists/listinfo/tmda-users
