<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.26.0">
</HEAD>
<BODY>
Thanks for that, Downloaded and imported here. <BR>
<TABLE CELLSPACING="0" CELLPADDING="0" WIDTH="100%">
<TR>
<TD>
<BR>
</TD>
</TR>
</TABLE>
<BR>
-----Original Message-----<BR>
<B>From</B>: Steve Freegard <<A HREF="mailto:Steve%20Freegard%20%3csteve.freegard@fsl.com%3e">steve.freegard@fsl.com</A>><BR>
<B>Reply-to</B>: MailScanner discussion <mailscanner@lists.mailscanner.info><BR>
<B>To</B>: MailScanner discussion <<A HREF="mailto:MailScanner%20discussion%20%3cmailscanner@lists.mailscanner.info%3e">mailscanner@lists.mailscanner.info</A>><BR>
<B>Subject</B>: 419 Spams<BR>
<B>Date</B>: Sun, 18 Oct 2009 22:35:40 +0100<BR>
<BR>
<PRE>
Hi all,
I have access to a system that receives a *lot* of 419-type spam
e-mails, so using an SA plug-in that I wrote recently (SaveHits:
<A HREF="http://www.fsl.com/support/SaveHits.pm">http://www.fsl.com/support/SaveHits.pm</A>); I've captured over 5000 since
15th October.
These came in very useful recently for me to increase the accuracy of
several bayes databases that were not accurately catching 419s so I've
made it available for others that might find this useful as well.
I've obfuscated the e-mail addresses, domains and source IP address
within all of the messages so the originating site cannot be identified.
You can download it from <A HREF="http://www.fsl.com/support/419_spams_1009.tar.gz">www.fsl.com/support/419_spams_1009.tar.gz</A> and
import it into your bayes database by running:
tar -zxf 419_spams_1009.tar.gz
sa-learn --spam --dir 419_spams
Obviously - this won't help if the bayes database has been incorrectly
trained for a considerable amount of time but worked for me when
starting afresh and letting bayes autolearn 200 ham messages from the
actual mail stream.
Kind regards,
Steve.
</PRE>
</BODY>
</HTML>