[SAtalk] preventing blacklistes messages from impacting bayesian learning?

klists and whitelists get ignored by default with a recent version? I found some emails in the archive which seemed to imply that they don't, and that they cause problems for the learner. Thanks, Britton Kerin __ GNU GPL: "The Source will b

[SAtalk] send RAW or SHOWN text from pine to sa-learn?

Under pine, I first turn full headers on, then use '|' to pipe message to sa-learn --(spam or ham) --no-rebuild --single. Until now, I have been using the SHOWN text, but I just noticed that pine can also send RAW text. Which should be used? Britton Kerin __ GNU GPL: "The Sour

Re: [SAtalk] Re: False Positive (non mailing list)

> I didn't see what SA had done with it or the tests or scores Okay, here's SA's analysis: X-Spam-Status: Yes, hits=6.5 required=5.0 tests=DOUBLE_CAPSWORD,CALL_FREE,LINES_OF_YELLING, LINES_OF_YELLING_3,LINES_OF_YELLING_2,SUPERLONG_LINE, BIG_FONT,PORN_3,MSG_ID_

[SAtalk] False Positive (non mailing list)

In the many months I've been using SpamAssassin, I've only seen one false positive. I just checked it against the latest SA and it still gets flagged. I wonder if this could have been avoided (other than by raising the threshold, of course)? Suzanne - --- >From tril We