You man want to take a look at spambayes. Yes it is still clasified as
alpha but
I have been using it successfuly on my server for couple of months.
Just set it
up in a prxy mode and train it. Works like a charm.
Greg
p.s.
It is pretty easy to set up postfix to listen on more then one port
therefore you could
send mail through it even with port 25 blocked.
der.hans wrote:
>Am 17. Jun, 2003 schwätzte Michael Sheldon so:
>
>
>
>>Spamassasin is nice, but does have a significant false-positive rate. Since
>>it primarily filters on content, most legit commercial email (Order
>>receipts, order status messages, commercial mailing lists, etc.) get
>>"whacked" by it.
>>
>>
>
>Make it easy for your users to add incoming addies to white lists.
>
>Also, I've set my filter to be 10 and believe I've only had one false
>positive. I believe it was a forwarded message in Polish and the guy who
>sent it to me is likely to have sent a either a joke that would set off
>triggers or sent me some sort of warning that would include SPAM or
>SPAM-like content.
>
>A lot has still been getting through at 10, but I'm hoping the bayesian
>filter will start catching more of that.
>
>
>
>>This means it really can only be used as a "tagging" agent, not as a means
>>to reject or discard mail. Which of course, means that you have to accept,
>>then hand filter through the trash. I love it for it's ability to prioritize
>>things for me, but it's not a true solution for spam.
>>
>>
>
>A suggestion from Brian was to put the SPAM ranking in the subject, then
>sort the SPAM folder by subject. This allows you to look at items based on
>their SPAM ranking. 48 is the highest ranking I've gotten thus far. Almost
>half my tagged SPAM was ranked over 15.
>
>subject_tag SPAM(_HITS_):
>
>ciao,
>
>der.hans
>
>
--
"Anger and Greed are the primary causes of defeat"
Sun Tzu - "Art of War"
(text/plain)