The same goes for me for mortgages.. i dont have one and dont need one. so anything about mortgages, home financing etc is definitely spam for me.
Ahh, but do you want to automatically declare any email which mentions such things to be spam?
Do you have absolutely no financial accounts of any sort whatsoever that would ever contact you by email?
Take into consideration an email from your bank which largely discusses matters regarding new options available in checking accounts, but might also mention mortgages.
Perhaps you do some stock trading, even as a simple 401k.. maybe you might get savy and subscribe to the motley fool or some other newsletter.. Clearly it's not inappropriate for these newsletters to mention non-stock financial matters in passing. "Low mortgage rates have recently been inflating the real estate market, and xxx fund has been reaping the benefits by investing heavily in real estate related companies".
It's very easy to make generalizations which later turn out to be grossly invalid because you were overly narrow in your thinking, so I'd caution you against making such broad generalizations as "anything to me involving mortgages is spam".
It's not outright invalid to do things like this, but be very wary as you might just get what you ask for, instead of what you want.
So my initial reaction to this solution is to allow per-user content filtering by doubling the points for each test. In the grandmother example, i have a list of about 20 tests, and i'm doubling the points earned for each of those tests so that any porn will be tagged/flagged as spam, but all other kinds of email will be filtered using the default rules.
Doubling is a bit severe, but the concept is somewhat valid.. However, I personally feel that "per user" learning is really best handled by having per-user bayes databases. Score tweaking can't even get remotely close to what you can achieve by feeding independent bayes databases per-user...
Is there anything fundamentally flawed with my idea? Would this really mess up the filtering mechanism?
No, but I think other methods would work significantly better.
------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. Does SourceForge.net help you be more productive? Does it help you create better code? SHARE THE LOVE, and help us help YOU! Click Here: http://sourceforge.net/donate/ _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
