2.9.8 Not working so well

SpamSieve
1

Since I upgraded to 2.9.8, I’ve been getting many penis enlargement ads. Never got them before, but now they’re non-stop. Help?

2

Did you make any other changes, such as updating your operating system?

In any case, it sounds like there’s a setup problem. Please follow the instructions on the Why is SpamSieve not catching my spam? page.

3

I’ve re-installed and everything. I’ve got some log info that might help? Thanks…

Predicted: Good (7)
Subject: Be the Pied Piper of chicks
From: barometerfloc@dvb-brasil.org
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Reason: P(spam)=0.001[0.064], bias=1.000, F:enlargement(0.999), S:chicks(0.057), H:Content-Language(0.071), XM:MicrosoftOfficeOutlook12.0(0.079), ^fg-0000ff(0.082), C:Math(0.091), content-language:en-us(0.099), C:Cambria(0.102), ^a-class-MsoNormal(0.109), ^a-class-MsoNormal(0.109), ^a-class-Section1(0.118), F:Sample(0.878), C:1026(0.127), C:xml(0.128), C:xml(0.128)
Date: 2013-09-30 11:40:25 -0700

Trained: Good (Auto)
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Actions: added rule <From (address) Is Equal to “barometerfloc@dvb-brasil.org”> to SpamSieve whitelist, added rule <From (name) Is Equal to “Sample enlargement”> to SpamSieve whitelist
Date: 2013-09-30 11:40:26 -0700

Trained: Spam (Manual)
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Actions: disabled rule <From (address) Is Equal to “barometerfloc@dvb-brasil.org”> in SpamSieve whitelist, disabled rule <From (name) Is Equal to “Sample enlargement”> in SpamSieve whitelist, added rule <From (address) Is Equal to “barometerfloc@dvb-brasil.org”> to SpamSieve blocklist, added rule <From (name) Is Equal to “Sample enlargement”> to SpamSieve blocklist, added to Spam corpus (1375)
Date: 2013-09-30 12:37:30 -0700

Mistake: False Negative
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Classifier: Bayesian
Score: 7
Date: 2013-09-30 12:37:35 -0700

Trained: Good (Auto)
Subject: So hard you can break an egg
Identifier: hUSH9ml9Ubso4Y9k4S3ZFA==
Actions: added rule <From (address) Is Equal to “unidirectionalallegra@partenaire-entreprise.fr”> to SpamSieve whitelist, added rule <From (name) Is Equal to “Sample Men’s Supplement”> to SpamSieve whitelist, added to Good corpus (902)
Date: 2013-09-29 22:27:43 -0700

4

I’ve re-installed…
I’ve re-installed (before I posted the first time). Perhaps this log entry will help you? I’m not really smart enough, I’m afraid…

=====================================================================
Predicted: Good (7)
Subject: Be the Pied Piper of chicks
From: barometerfloc@dvb-brasil.org
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Reason: P(spam)=0.001[0.064], bias=1.000, F:enlargement(0.999), S:chicks(0.057), H:Content-Language(0.071), XM:MicrosoftOfficeOutlook12.0(0.079), ^fg-0000ff(0.082), C:Math(0.091), content-language:en-us(0.099), C:Cambria(0.102), ^a-class-MsoNormal(0.109), ^a-class-MsoNormal(0.109), ^a-class-Section1(0.118), F:Sample(0.878), C:1026(0.127), C:xml(0.128), C:xml(0.128)
Date: 2013-09-30 11:40:25 -0700

Trained: Good (Auto)
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Actions: added rule <From (address) Is Equal to "barometerfloc@dvb-brasil.org"> to SpamSieve whitelist, added rule <From (name) Is Equal to “Sample enlargement”> to SpamSieve whitelist
Date: 2013-09-30 11:40:26 -0700

Trained: Spam (Manual)
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Actions: disabled rule <From (address) Is Equal to "barometerfloc@dvb-brasil.org"> in SpamSieve whitelist, disabled rule <From (name) Is Equal to “Sample enlargement”> in SpamSieve whitelist, added rule <From (address) Is Equal to "barometerfloc@dvb-brasil.org"> to SpamSieve blocklist, added rule <From (name) Is Equal to “Sample enlargement”> to SpamSieve blocklist, added to Spam corpus (1375)
Date: 2013-09-30 12:37:30 -0700

Mistake: False Negative
Subject: Be the Pied Piper of chicks
Identifier: DWBuHlDdpdebnqSfeAS5xg==
Classifier: Bayesian
Score: 7
Date: 2013-09-30 12:37:35 -0700

5

It looks like you have the Spam-catching Strategy set to the most aggressive, which is probably not a good idea because it can lead to false positives. My guess is that there’s something wrong with the training so that this message wasn’t caught. If you e-mail me your log file I can better advise you.

6

Thanks for sending the log file. It looks like some of the mistakes went uncorrected, and others were corrected days late. It’s best to correct all the mistakes promptly. If that’s not possible, you should turn off auto-training to prevent one mistake from leading to a cascade of others. At this point, it looks like SpamSieve is working from some incorrect and outdated information about which e-mails are spam, so I suggest resetting the corpus and re-training it.