Hi Andrey,
 
we've implemented such a test in SpamChk, an external Test for Declude
Junkmail. There are two different test for less and more then 10 spaces.
Some spaces can occur sometime in real messages, but spammers try to hide a
ID at the end of the subject line so if there are (a lot) more then 10
following spaces its a good test for spam.
 
In any case you shouldn't take any action only for a single spam attribute
like this. In a weighting system as used in Declude Junkmail a space-test
should have a weight between 30 and 60% but not more to avoid false
positives.
 
Keep in mind that with such a rule you can't catch more then 5% of all spam
messages having on the other side some false positives. As part of a good
balanced weighting system using over 50 DNS- RFC- and content-based tests
the effectiveness can be increased up to 90 - 95%.
 
Declude Junkmail costs something (not to much for it's functionality and
support) 
SpamChk can be used with Junkmail and it's free.
 
Markus
 
 

-----Original Message-----
From: Andrey Fyodorov [mailto:[EMAIL PROTECTED]] On Behalf
Of Andrey Fyodorov
Sent: Tuesday, January 21, 2003 2:49 AM
To: [EMAIL PROTECTED]
Subject: lost of spaces in the subject


Hey all. I am trying to create a primitive spam filter. This one will
specifically look at the subjects of messages.
 
Lots of spam mail has spaces in the subject followed by some gibberish. So I
have created a rule that would check the subject for 10 spaces in a row
followed by something else.
 
It is working really well, except I suspect that it trapped a few non-spam
messages.
 
What kind of programs are out there that would produce non-spam messages
with lots of spaces in the subject line?

<<attachment: winmail.dat>>

Reply via email to