The problem is for this combination of 'spam like features' your training
samples have '8' non spam messages, and '0' spam messages.
So essentially, you've trained it that in most cases, these features mean
the message is not spam :-)
Assuming the training has not been deliberately mixed up. Then the
problem is actually that the detection criteria are too general and training
isn't going to improve results it's just going to change the errors from false
one way to false the other way as it keeps trying to learn two meanings for
'one' thing.
ChrisP.
Huh, The quarantine page says "0", but the header has a score of "1."
It doesn't matter. One is way too low.
Here is another one I got today.
Received-SPF: pass (Last token {ip4:89.43.11.72} (res=PASS))
client-ip=89.43.11.72;
envelope-from=<CarlosLeonard@esg-kuwait.yoncoo.com>;
x-ip-name=esg-kuwait.yoncoo.com;
Received: from esg-kuwait.yoncoo.com (unverified [89.43.11.72])
by ericvey.com (SurgeMail 6.6a) with ESMTP id 14268-1312327
for <junker@ericvey.com>; Tue, 24 Jun 2014 14:39:48 -0400
Return-Path: <CarlosLeonard@esg-kuwait.yoncoo.com>
X-Verify-SMTP: Host 89.43.11.72 sending to us was not listening
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
To: <junker@ericvey.com>
Reply-to: <Carlos.Leonard.reply@yoncoo.com>
Subject: Hi, we would like to thank you ($1,000-Card)
Message-ID: <Carlos.20140624015257.28948.58996.junker@ericvey.com>
Date: Tue, 24 Jun 2014 11:39:17 -0700
From: Sears Dept.-Store <Carlos@esg-kuwait.yoncoo.com>
X-Original-MessageID: <bfb3f280b4369095133f3d893eb946fa>
X-Originating-IP: 89.43.11.72
X-Country: code=RO country="Romania" ip=89.43.11.72
X-Rcpt-To: <junker@ericvey.com>
X-Vpipe: Scanner said ok (av_avast)
X-SpamDetect: : 0.0 sd=0.0 Exact nspam=0 nok=1 0.00 $0.94(SpamUrl)
$0.85(X-Verify-SMTP present) $0.80(X-Phrase:isspam)
$0.70(X-myrbl:unknown) $0.68(SPF Negate) $0.40(X-NotAscii:utf)
$0.42(X-LangGuess:English) $0.56(X-SpamUrl:_EXISTS_) $0.56(SpamUrl)
$0.45(isclickimage1)
X-NotAscii: charset=utf-8
X-SpamUrl: sears.com
X-LangGuess: English
X-Probe: +OK nothing bad found
X-Phrase: IsSpam score=1.00
X-MyRbl: Color=Unknown ip=89.43.11.72
X-IP-stats: No info recorded yet ip=89.43.11.72
------ Original Message ------
From: "Eric Vey" <junker@ericvey.com>
To: surgemail-list@netwinsite.com
Sent: 6/24/2014 4:19:00 PM
Subject: [SurgeMail List] Amazing amount of spam scores 0 points
Hello,
Last week I changed a couple of accounts (including this one since this
list is published and indexed by google) to quarantine anything that
had a smite score of 0 or above. A few legitimate messages from other
lists were caught, but many more were spam messages that got a score of
0.
I have been religiously marking spam in the quarantine page and
forwarding spam that still got through to isspam at my domain.
Here are the headers of one I got today.
Received-SPF: pass (Last token {mx} (res=PASS))
client-ip=148.251.72.78;
envelope-from=<WhosWho-junker=ericvey.com@gcatri.com>;
x-ip-name=vps.gcatri.com;
Received: from vps.gcatri.com (unverified [148.251.72.78])
by ericvey.com (SurgeMail 6.6a) with ESMTP id 14290-1312327
for <junker@ericvey.com>; Tue, 24 Jun 2014 15:48:53 -0400
Return-Path: <WhosWho-junker=ericvey.com@gcatri.com>
X-Verify-SMTP: Host 148.251.72.78 sending to us was not listening
DKIM-Signature: v=1; a=rsa-sha1; c=relaxed/relaxed; s=dkim;
d=gcatri.com;
h=Mime-Version:Content-Type:Message-Id:Date:From:To:Subject;
i=WhosWho@gcatri.com;
bh=p0EHgg3YNLZjMDWxc9elw5EvcK0=;
b=EcELmJqBQyWaDbhqS10m0oIqXZjyQvMSDCqlJYS9KhKmaxZ0FIv3n11VUOM+GmwsiYowVqmqn9pi
+Uagvc01u9hZ84baskUaUH+hLETR9knSuVU0RUBMkfJYmu4sE8Qddbppond6pXNc/bMlKLAfK/9e
JDB7DrOSnlOO7krvFnM=
Received: by vps.gcatri.com id hl7ats0001gp for
<junker@ericvey.com>;
Tue, 24 Jun 2014 19:47:34 +0000 (envelope-from
<WhosWho-junker=ericvey.com@gcatri.com>)
Mime-Version: 1.0
Message-Id:
<29935be4871fdf37a0211382485d9c02.8d4a0f37918d02b6@gcatri.com>
Date: Tue, 24 Jun 2014 19:47:34 +0000
From: Whos Who <WhosWho@gcatri.com>
To: junker@ericvey.com
Subject:
=?utf-8?B?Q29uZ3JhdHVsYXRpb25zISBZb3UncmUgYSAyMDE0IENhbmRpZGF0ZSBmb3IgV2hvJ3MgV2hvISBDb25maXJtIE5vdy4=?=
X-Originating-IP: 148.251.72.78
X-Country: code=DE country="Germany" ip=148.251.72.78
X-Rcpt-To: <junker@ericvey.com>
X-Vpipe: Scanner said ok (av_avast)
X-SpamDetect: : 0.0 sd=0.0 Close nspam=0 nok=8 0.00
$0.96(X-LangGuess:Wrong) 0.94(SpamUrl) $0.85(X-Verify-SMTP present)
0.81(unknown_lang) 0.80(X-Phrase:isspam) $0.21(genuine)
0.70(X-myrbl:unknown) 0.68(SPF Negate) 0.40(X-NotAscii:utf)
0.56(X-SpamUrl:_EXISTS_) 0.56(SpamUrl)
X-NotAscii: charset=utf-8
X-SpamUrl: gcatri.com
X-LangGuess: Unknown, Wrong
X-Probe: +OK nothing bad found
X-Phrase: IsSpam score=1.00
Authentication-Results: ericvey.com
header.from=WhosWho-junker=ericvey.com@gcatri.com; dkim=pass (good
signature)
X-MyRbl: Color=Unknown ip=148.251.72.78
X-IP-stats: No info recorded yet ip=148.251.72.78