This is probably an easy one, but I'm not able to find anything to
explain the contradiction in the docs.

The auto_learn_threshold_nonspam and auto_learn_threshold_spam configs
are at the defaults (at least I haven't affected them).  However, I'm
still getting messages scoring well above the
auto_learn_threshold_spam setting of 15 that aren't being autolearned.

Here are the spamassassin headers from one such message:

X-Spam-Status: Yes, hits=24.0 required=6.0
        tests=BODY_8BITS,CHARSET_FARAWAY,CLICK_BELOW,FROM_ENDS_IN_NUMS,
              HEADER_8BITS,HTML_60_70,HTML_FONT_BIG,HTML_FONT_COLOR_BLUE,
              HTML_FONT_COLOR_GREEN,HTML_FONT_COLOR_NAME,
              HTML_FONT_COLOR_RED,HTML_FONT_COLOR_UNSAFE,
              HTML_FONT_FACE_BAD,HTML_FONT_FACE_ODD,HTML_MESSAGE,
              KOREAN_UCE_SUBJECT,MAILTO_TO_SPAM_ADDR,MANY_EXCLAMATIONS,
              MIME_HTML_ONLY,MSG_ID_ADDED_BY_MTA_3,NO_REAL_NAME,
              SUBJ_FULL_OF_8BITS,UNDESIRED_LANGUAGE_BODY
        version=2.50
X-Spam-Level: ************************
X-Spam-Report:   ---- Start SpamAssassin results
  24.00 points, 6 required;
  *  4.3 -- Subject: contains Korean unsolicited email tag
  *  0.7 -- From: ends in numbers
  *  1.2 -- Headers include 3 consecutive 8-bit characters
  *  0.7 -- From: does not include a real name
  *  0.2 -- BODY: HTML font color has unusual name
  *  0.2 -- BODY: HTML font face is not a commonly used face
  *  0.1 -- BODY: Message is 60% to 70% HTML
  *  0.1 -- BODY: FONT Size +2 and up or 3 and up
  *  0.1 -- BODY: HTML font color is red
  *  0.1 -- BODY: HTML font color not within safe 6x6x6 palette
  *  0.1 -- BODY: HTML font color is blue
  *  4.0 -- BODY: Written in an undesired language
  *  3.2 -- BODY: Character set indicates a foreign language
  *  1.5 -- BODY: Body includes 8 consecutive 8-bit characters
  *  0.7 -- BODY: HTML font color is green
  *  0.7 -- BODY: HTML font face is not a word
  *  0.1 -- BODY: HTML included in message
  *  0.4 -- URI: Includes a link to a likely spammer email address
  *  4.3 -- Subject is full of 8-bit characters
  *  0.3 -- 'Message-Id' was added by a relay (3)
  *  0.1 -- Asks you to click below
  *  0.1 -- Message only has text/html MIME parts
  *  0.8 -- Subject has many exclamations
  ---- End of SpamAssassin results

I do have some with lower scores that ARE being autolearned.  Here's
one example:
X-Spam-Status: Yes, hits=17.7 required=6.0
        tests=HEADER_8BITS,HTML_70_80,HTML_FONT_BIG_B,
              HTML_FONT_COLOR_BLUE,HTML_FONT_COLOR_NAME,
              HTML_FONT_FACE_BAD,HTML_FONT_FACE_ODD,HTML_IMAGE_ONLY_06,
              HTML_MESSAGE,KOREAN_UCE_SUBJECT,MAILTO_TO_SPAM_ADDR,
              MIME_HTML_ONLY,MSG_ID_ADDED_BY_MTA_3,RAZOR2_CHECK,
              RCVD_IN_RFCI,SUBJ_FULL_OF_8BITS
        autolearn=spam  version=2.50
X-Spam-Level: *****************
X-Spam-Report:   ---- Start SpamAssassin results
  17.70 points, 6 required;
  *  4.3 -- Subject: contains Korean unsolicited email tag
  *  1.2 -- Headers include 3 consecutive 8-bit characters
  *  0.2 -- BODY: HTML font color has unusual name
  *  0.2 -- BODY: HTML font face is not a commonly used face
  *  0.5 -- BODY: HTML has a big "font" and "B" tag combo
  *  0.4 -- BODY: Message is 70% to 80% HTML
  *  0.1 -- BODY: HTML font color is blue
  *  1.1 -- BODY: HTML has images with 400-600 bytes of words
  *  0.7 -- BODY: HTML font face is not a word
  *  0.1 -- BODY: HTML included in message
  *  0.4 -- URI: Includes a link to a likely spammer email address
  *  2.0 -- Listed in Razor2, see http://razor.sf.net/
  *  4.3 -- Subject is full of 8-bit characters
  *  0.3 -- 'Message-Id' was added by a relay (3)
  *  1.8 -- RBL: Received via a relay in ipwhois.rfc-ignorant.org
  [RBL check: found 57.24.187.211.ipwhois.rfc-ignorant.org., type: 127.0.0.6]
  *  0.1 -- Message only has text/html MIME parts
  ---- End of SpamAssassin results

Any idea why that one didn't get autolearned?  I assume the answer is
somewhere in these headers, but I can't see it.

TIA
Lou
-- 
Louis LeBlanc               [EMAIL PROTECTED]
Fully Funded Hobbyist, KeySlapper Extrordinaire :)
http://www.keyslapper.org                     ԿԬ

Experiments must be reproducible; they should all fail in the same way.


-------------------------------------------------------
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

Reply via email to