[Bioperl-l] killing spam with message body matching

Chris Dagdigian dag@sonsorol.org
Fri, 9 Mar 2001 12:47:49 -0500 (EST)

This is totally off topic for the list but since you asked and it
relates to spam leaking on to bioperl-l....

I've found that the only reasonable way to block inbound spam is at the
sendmail/qmail level via the use of DNS-based IP lookups on block lists
maintained by the kind folks at mail-abuse.org

The bioperl.org server currently makes use of 2 such services, The RBL
(reltime blackhole list) and RSS (relay spam stopper). The RBL is a list
of IP addresses of known spammers and the RSS is a frequently-updated
list of internet mailservers that allow unrestricted SMTP relay. Every
single inbound SMTP connection to our project is validated against both
of these lists before it is accepted.

The combination of RBL and RSS on the bioperl.org server has stoped spam
dead to the point where we now get about 2-4 messages *per year* leaking
on to our lists. I have yet to find anything that is more effective.


On Fri, 9 Mar 2001, Eugene Leitl wrote:

> As you probably noticed, spammers modify message bodies with random
> strings, so a simple cryptographic hash on message body won't do for spam
> filtering.
> Which matching algorithm would you suggest, which would reliably recognize
> randomly tagged spam?
> TIA,
> -- Eugene

Chris Dagdigian         (Home:Work) Blackstone Technology Group
dag@sonsorol.org             :      dagdigian@ComputeFarm.com
http://www.sonsorol.org      :      http://www.computefarm.com
http://open-bio.org          :      Mobile (617) 877-5498
Schedule & full contact info http://www.sonsorol.org/dag/contact.html