I've been working on a mail filter that is strictly per-user configured.
It's just a Bayes filter modeled after bogofilter but it uses a
database backend and plugs into the smtp mail stream (so it can be
run on a mail server or proxy).
It's a fun project I've been using for about a year.
Haven't really measured the accuracy in a while.
There are some parameter tuning I could do but I haven't found the
time for that just yet.
but >95% is a certainty, I'll even venture to 99% but I'm not sure
about decimal points after that.
It falls into a "good enough" for me and it's pretty brain dead
simple to manage/use.
Love the postgresql version 8.2 -- couldn't do it without that.
On Nov 5, 2007, at 6:52 AM, Charles Marcus wrote:
On 11/4/2007, Tom Allison ([EMAIL PROTECTED]) wrote:
Dspam isn't something I would ever consider using again. After
several postings to the group with specific logs showing the lost
messages, and never receiving acknowledgment that there might be
any kind of a bug. Dspam is not for me.
The main problem with DSPAM is it is not very easy to configure
correctly, or to optimize.
ASSP, on the other hand, works out of the box, with *very* little
config necessary, and the defaults and default spam corpus provide
an out of the box accuracy of 95+%...
With a bit of training, it quickly gets up to 99+%.
As I said - the only downside to ASSP is it doesn't provide per-
domain or per-user settings - although there is a commercial cPanel
version that does, if you happen to be running a cPanel server.
I'd sure like to see similar functionality provided by webmin/
virtualmin - even if it was limited to Virtualmin Pro.
--
Best regards,
Charles
_______________________________________________
DBmail mailing list
[email protected]
https://mailman.fastxs.nl/mailman/listinfo/dbmail
_______________________________________________
DBmail mailing list
[email protected]
https://mailman.fastxs.nl/mailman/listinfo/dbmail
