> -----Original Message----- > From: Raul Dias [mailto:[EMAIL PROTECTED] > Sent: Thursday, July 10, 2003 4:03 PM > To: [EMAIL PROTECTED] > Subject: [SAtalk] How to balance new rules for hams that > looks like spam > > > Hi, > > > Now I am facing a problem with hams. > > Some advertising mails are legitimate mails that the user did > subscribe to receive it. > The problem is that it looks like spam, uses spams techniques > (like "web_bugs"), it smells like spam. But it isn't. > > Some of this mails are getting even medium scores like > 10 to 15 points. > Some get BAYES_90 or BAYES_99 scores. > To make things worse a few is listed in some open RBLs. > > I donīt know the inners of the Bayesian database, but I think that > if I fed it as ham to the database I would make the bayesian filter > less efficient. Is this true? > > So, how do you balance this kind of situation? > Is whitelisting the only way to go? > > Regards, > > Raul Dias > > >
I had some simialr problems. I concentrated on CNET news email. Looks totally spammy. I tried to work it out without whitelisting. Conclusion.......Whitelist it :) If you know a certain person or group likes a particular kind of newsletter, say tennis, you could write some negative rules for phrases specific to Tennis. This works pretty good. Unless your users are into enlarging their body parts and would like to make money from home ;) Chris ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps1 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
