On Thu, 2016-09-29 at 20:28 -0400, Alex wrote:
> 
> Hi all,
> 
> Has anyone given any thought to special rules or methods designed to
> catch greymail? That is, mail that perhaps may be opt-in, but
> abusive,
> like marketing mailing lists or newsletters?
> 
I use metas for this sort of unwanted mail, in combination with my
portmanteau rule builder. What I refer to as a 'portmanteau rule' is
one with a large list of alternates.

For example, to catch sales junk I use two portmanteau rules: one
contains selling terms and phrases I've found in junk mails and the
other rule has product and brand names, also harvested from sales spam.
Both rules score very low, 0.01 or less, while the 'sales spam' meta
rule has a relatively high score and is only triggered if both the
portmanteau rules have fired. One benefit of this approach is that it
also catches combinations that haven't previously been used by spammers
while, because any of these combinations are rare in ham, generates
almost no FPs provided you're reasonably careful about what terms and
phrases you include in your portmanteau rules.

I wrote the portmanteau rule builder to make editing these giant
alternate list rules easier: the rule definition file has each
alternate regex on a separate line. It is easily portable because its
just an awk script embedded in an bash script.

Full documentation and code is here: 
http://www.libelle-systems.com/free/portmanteau/portmanteau.tgz


Martin

Reply via email to