warren, anthony wrote:
> Hi all,
>
> After some testing with SMTP sender verify callouts and getting mixed
> results from that method, I am keen to explore other Spam reduction
> techniques.
spam filtering techniques are like underwear. What fits one person is
often an automatic wedgie for others.
After getting hurt by a bunch of different simpleminded anti-spam
techniques, I created a hybrid technique based on proof of work puzzles,
automatic white lists, and as a fullback, content filtering. The
project is called camram or campaign for real mail (no, I did not name
it). You can get the sales pitch at
www.camram.org.
in a nutshell, I found that using a variety of techniques which allow
you to extract e-mail from known good sources without requiring the
average user to do a dam thing improves filtering characteristics
significantly and eliminates false positives from regular e-mail sources.
at the end of the day, when everything is working right, you have an
anti-spam system that requires very little user involvement, and easy to
manage spamtrap, the dumpster where it's easy to recover false positives
and a host of other really important user interface features. I cannot
emphasize us enough. It doesn't matter how good the technique is, if
the user interface is painful to use (or a system administration jobs
program), it's not doing your users any favors to implement it.
one of the problems on wrestling with right now is, it really works best
when called by a local delivery agent. But when running in a server, it
really should be called as a filter when the message has been queued but
not delivered. Additionally, it's filtering messages both ways.
Inbound filtering for content, outbound, adding stamps.
I'm currently looking at both exim and postfix on ways to do this. I
will admit the postfix has the upper hand with its built-in smtp to
filter protocol. the python extension to exim is nice but I'm not sure
it fills my needs entirely.
---eric
--
Speech recognition in use. It makes mistakes, I correct most
