Hello Mike,.

MK> For myself...I will make it so a single line of "PEN?S
MK> ENLARGEMENT" is enough to trigger as spam.

The value of SA is that it can be customized for individual
user needs, and that it is easy to change scoring, as well
as to write your own rules.

I also have some words and phrases that I feel should always
be treated as spam. I personally have chosen to use SA in
conjunction with procmail, and I have written procmail
recipes which automatically delete or divert such emails
even before SA is invoked; my logic is simply that if a
single test is going to be determinative, there is no need to
waste server resources running a series of tests to score
the email. (I'm managing a very busy server so I have to be very
aware of memory load issues).

I think that as you use Spamassassin you will get a better
sense of what it can do, and how best to adjust it to your
needs.  For example, if phrases like "Amazing sex" or
"Climatique" appear in the subject line of any email, the
email gets deleted without ever being scored. (I also have
procmail send a copy to a file that is fed daily to
sa-learn, so I don't lose the benefit of having these spams
included in the database).

I realize it is frustrating at first to have to do so much
fine tuning, but it really does pay off in the end. To me,
the value of SA is not that it can catch bad words and
phrases that I already have encountered -- it is easy for me
to write simple filters for that -- but that the combination
of weighting and use of Bayes filtering is able to catch
most new variations as they come through.

-Abigail



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to