Thus the difference between the two runs is:
3.8 + BAYES_60 + NO_REAL_NAME = 6.0
3.8 + 1.2 +1.0 = 6.0
At 11:07 AM 3/18/2003 +0000, Darren Coleman wrote:
Hi,
I received two identical emails to two different email addresses in the space of a few minutes that are both covered by a single instance of SpamAssassin, and one of them was deemed to be spam, and the other not. This in it of itself wouldn't be a massive issue were it not for the fact that the scores do not tally up.
(I have attached both emails to this message)
The first mail (ham), the SA test matches returned were:
X-Spam-Status: No, hits=3.8 required=5.0 tests=FROM_HAS_UNDERLINE_NUMS,HTML_70_80,HTML_IMAGE_ONLY_02, HTML_IMAGE_RATIO_02,HTML_WITH_BGCOLOR,MIME_HTML_ONLY, PRIORITY_NO_NAME version=2.50
The second mail (spam) had the following SA test matches:
X-Spam-Status: Yes, hits=6.0 required=5.0 tests=BAYES_60,FROM_HAS_UNDERLINE_NUMS,HTML_70_80, HTML_IMAGE_ONLY_02,HTML_IMAGE_RATIO_02,HTML_WITH_BGCOLOR, MIME_HTML_ONLY,NO_REAL_NAME,PRIORITY_NO_NAME version=2.50
Aside from the sender name, both emails appear are 100% identical, and bave the exact same SA matches except for the BAYES_60 (1.2 points) match in the spam-tagged mail. I can only presume that the Bayesian filter "learnt" something from the "ham-spam" email, which is a "good thing".
Nothing wrong here, except for the fact that 3.8 + 1.2 != 6.0.
So, can anyone see what's gone wrong here?
-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
