1 |
* On Sep 26 14:57, Jose Gonzalez Gomez (gentoo-server@l.g.o) wrote: |
2 |
> As far as I know about DSPAM (anybody correct me if I'm wrong), you |
3 |
> must train it in order to begin detecting spam. I have never |
4 |
> configured it, but I'm using it, and it provides two special addresses |
5 |
> per user: spam-user@××××××××××.com and fp-user@××××××××××.com. |
6 |
> Whenever you receive spam not detected by DSPAM you must forward it to |
7 |
> the first address. Whenever DSPAM incorrectly tags a mail as SPAM you |
8 |
> must forward it to the second address (false positive). The more you |
9 |
> train it, the more accurate it gets. |
10 |
|
11 |
I trained DSPAM for over 6 months. It never gained accuracy past the |
12 |
first week, topping off at about 93%. (And yes, I did check that it was |
13 |
still learning, picking up new tokens, etc.) |
14 |
Tom |