Gidday folks. I'm pretty sure my Bayes database is muntered, because an awful lot of ham is receiving a full Bayes penalty. It's been particularly hard keeping this list's posts from being tagged.
>From Pete Dubler: 0.999-2--0h-46s--0d--Indeed, ...meaning the word "Indeed" has been found in 46 spam but in NO ham messages? Am I reading that correctly? I've got tens of thousands of live accounts here and I would expect some of them to receive the word "Indeed" on occasion. Bayes auto learning is on, thresholds at -2 and +10. But I expect that problem will go away after a bit of sa-learning, starting with --clear. On to my question: Here are some Bayes spammytags on a recent post from Matt Kettler, subject "Re: some mail still not getting X-SPAM headers": 1.000-22--0h-606s--0d--readd, 1.000-4--0h-93s--0d--H*M:inc, 1.000-166--0h-4623s--0d--HReceived-SPF:designates, 1: "readd" wasn't in the message. Huh? 2: What is H*M:inc? "inc" was in the message ID? Where can I find an explanation of the colon-syntax? 3: What do you use for bayes_ignore_header? I have listed three that Gnus adds, but otherwise I'm relying on defaults. What's accepted practice? Many thanks in advance. -- _________________________________________________________________________ Andrew Donkin Waikato University, Hamilton, New Zealand
