C-Command Software Forum

Spam Sieve missing lots of spam

Spam Sieve used to be great - now it’s failing.

Example:

This spam was referred to as a false negative - which is wrong. What do I do?

Trained: Spam (Manual)
Subject: Wassup orders Ready For 70% OFF?
Identifier: iO4d5JiG2v32wfw32i5rPA==
Actions: disabled rule <From (address) Is Equal to “orders@grnc.info”> in SpamSieve whitelist, disabled rule <List-Unsubscribe Is Equal to “<http://email.grnc.info/u?id=ab4a0b6168eb5e752504ceb60d26>”> in SpamSieve whitelist, added rule <From (address) Is Equal to “orders@grnc.info”> to SpamSieve blocklist, added to Spam corpus (2653)
Date: 2014-03-04 06:13:16 +0000

Mistake: False Negative
Subject: Wassup orders Ready For 70% OFF?
Identifier: iO4d5JiG2v32wfw32i5rPA==
Classifier: Bayesian
Score: 11
Date: 2014-03-04 06:13:21 +0000

Please send in your log file via e-mail so that I can look into this.

Thanks for sending the log file. It looks like there were a bunch of uncaught spam messages (over the past few weeks) that you never trained as spam. It is important to correct all the mistakes. Otherwise, SpamSieve will be working from incorrect information about which messages are spam and good, and this will reduce the filtering accuracy. At this point, I recommend that you reset the corpus and retrain SpamSieve so that it can get a fresh start.