I've been tinkering with ASSP for a while now and while initial testing
seemed positive, lately I'm easily seeing 40-60 emails per day bypassing
the spam filter.
When I've been trying to find out what goes wrong, I find that the
emails which don't get caught seem to be missing bayes and hmm
probability scores.
For example, this is one that got through:
X-Assp-ID: mail.sensenet.nu m1-60230-05081
X-Assp-Session: 7FC786895A60 (mail 1)
X-Assp-Envelope-From: ahlimajwv...@bosland.nl
X-Assp-Intended-For: h...@sensenet.nu
X-Assp-Version: 2.6.1(18128) on mail.sensenet.nu
X-Assp-Client-TLS: yes
X-Assp-Message-Score: 10 (PTR missing)
X-Assp-IP-Score: 10 (PTR missing)
X-Assp-Bayes-Confidence: 0.17441
X-Assp-HMM-Confidence: 0.63426
X-Assp-Spam-Level: ***
baysProbability is set to 0.6 and baysConf is 0.001, the "Add Bayes and
HMM Probability" and confidence headers are enabled. But I don't see
them here, even though the confidence is higher than 0.001
The emails that do get caught by the system are for the most part caught
because of all the other nice functions of ASSP, never by the bayes
filtering. When they do get caught, it's because HMM probability is 1.0.
It's almost as if it's not learning from the emails it receives and is
just acting on static information.
Maybe I don't understand fully how this part works, in that case, please
enlighten me :-) since the amount of spam bypassing ASSP is slowly
driving me mad.
/Henric
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Assp-user mailing list
Assp-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/assp-user
