Save me from spams

Scot L. Harris webid at cfl.rr.com
Wed Feb 9 18:49:57 UTC 2005


On Wed, 2005-02-09 at 11:53, arora.himanshu at gmail.com wrote:
> Hi Folks,
>              I am suffering from serious spam problem. I know that i
> can use spamassassin for this purpose but my system administrator has
> disabled this facility might be he thinks that it would slow down the
> server. I have no way to detect spams. I used procmail, it works for
> 20% of the cases because i used some specific dictionary words to
> check for spam because these mails are related to some particular
> subject. I cannot filter the mails because the mails are from
> different email IDs. The idea of using specific dictionary word failed
> because they used some special characters to represent some characters
> e.g. they used /\ for A. 1(one) or | (pipe) in place I( I for Indigo)
> \/ (backslash and forward slash ) for V.
>             I want a program which is not heavy in terms of CPU cycles
> consumed, which i can run from from my home directory(independent of
> binaries installed on the server). Might be a simple perl script. 
> Please reply me from where  i can get this program. It need not be
> very accurate. 70% to 80% accuracy is fine.
> 
> Himanshu Arora

Your email administrator should implement greylisting on the server. 
This will provide significant relief (possibly as much as 90 to 99%)
without imposing heavy CPU or bandwidth load on the server.  It is
available for most of the major MTAs out there (sendmail, postfix, etc).

If you can not get them to implement greylisting and/or spamassassin you
are left with trying to implement filtering in your account.  procmail
can be used for this.  Assuming for the moment that you are on the same
system as the MTA.  If not then you would need to use fetchmail then
procmail.  Have not done that before.  I have used filters built into a
client to invoke spamassassin and this works well.

I don't think you will have much luck with simple key phrase filtering. 
You will want to implement a bayes type filter that can learn and
adapt.  

The best option though is to get the email admin to do the right thing
and implement greylisting and/or spamassassin.  With both you can
achieve near 100% reduction in spam.


-- 

Response brought to you by AutoReponder 0.1
a product of Magic-8-ball productions.
(version 0.2 will feature correct answers!)





More information about the fedora-list mailing list