* On 22 Jul 2010, [email protected] wrote: 
> >  spam "X-Spam-Status: Yes, score=([^ ]+)" "SA:%1"
> >  # Add more patterns as needed for other spam engines you use.
>
> So, I'm experimenting with Bogofilter.  Bogofilter, doesn't seem to
> be catching much at all and has to be trained -- so I've been moving
> all spam to a folder and training it's filter.  However, so far, even
> though Bogofilter only brings in at most one dependency, it seems to
> be bogus at catching spam. :-/

I wasn't fond of bogofilter either, although having written my own
bayesian filter might have contributed to my lack of enthusiasm.  (Mine
was just an experiment and is long defunct.)  I may have tried dspam
afterwards.  It's been a while: I am now locked behind an "enterprise"
spam filtering system, and while I don't love it it's good enough.

I also like DCC for classification and I use a greylisting milter on my
home system.


> I guess instead of using a Bash script to call bogofilter to scan the
> spam maildir, I could simply create a hook to call it.  But I tend to
> try to do most things manually.

I found when I was doing this that invoking the filter manually was
too slow when I want to move on to the next message.  I saved it to a
mailbox and let a cron job eat the mailbox periodically.


> If I were using Spamassassin, I'd more then likely use this
> suggestion.

There's nothing peculiar to Spamassassin here.  I gave an example "spam"
rule for Spamassassin's headers, but any classifier can insert a header
and "spam" can match any header. :)

-- 
 -D.    [email protected]    IT Services    University of Chicago

Reply via email to