RE: [Exim] Spam protection from an outside server..

Tom Kistner Wrote:
>
> Kevin Reed wrote:
>
>
> > My idea of the system wide approach at least far as I've been
> > understanding is that SA is more built for user by user
> tuning rather
> > than as a whole system wide system at once. Could be that I'm just
> > missing something though...
>
> Not really. The bayesian part of SA learns from the
> "traditional" part, so it pretty much trains itself.

Tom,

The reason why I was thinking the way I was is based on statements like the
following in the documentation...

If you want to set up site-wide use of Bayesian classification, you should
set up a way for your users to send in misclassified mail to be "learned"
from. ...

...

Learning filters require training to be effective. If you don't train them,
they won't work. In addition, you need to train them with new messages
regularly to keep them up-to-date, or their data will become stale and
impact accuracy.


This led me to believe that without user participation in the learning
process... The ability to block spam would be not very effective.

> If you can impose a site-wide policy (for example, mark spam
> in headers and let users filter in Outlook), spamassassin is
> very nice to use, if you can provide the hardware needed for
> 8000-9000 users.

It was my original intention to mark and forward to let the users be able to
decide. Only the most obvious would be dropped if at all.

The user base is currently much higher but after the split it will be in
that range, maybe lower. All users have a standard desktop which currently
is based on a NT4 based system with Outlook etc.. About 2000 or so will be
offsite at various locations but will still be connected to the same
exchange servers as onsite. Outlook is '97.

Most of the users however that will not be part of the split will never be
effected by the new setup.

An upgrade of that standard will be occuring soon tis a pain not have USB
abilities.