18.04.2013 13:45, Joe Acquisto-j4 kirjoitti: >>>> On 4/18/2013 at 6:38 AM, Axb <axb.li...@gmail.com> wrote: >> On 04/18/2013 12:23 PM, Joe Acquisto-j4 wrote: >>>>>> On 4/18/2013 at 6:15 AM, Axb <axb.li...@gmail.com> wrote: >>>> On 04/18/2013 12:11 PM, Joe Acquisto-j4 wrote: >>>>> I'm missing something. >>>>> >>>>> Find a fair amount of missed SPAM showing, among others: >>>>> >>>>> * 0.8 BAYES_50 BODY: Bayes spam probability is 40 to 60% * >>>>> [score: 0.4968] >>>>> >>>>> Bayes is way too low, in my HO. >>>> it's obviously not learning enough of whatever it's not scoring higher... >>>> >>>>> I am puzzled by the line after it. >>>>> >>>>> I set local.cf with: score BAYES_50_BODY 3.6 and restarted sa. >>>>> >>>>> Still comes thru with the low score. Am I sticking it in the wrong >>>>> place? Or can it not be overridden? I do not recall if the odd >>>>> second line was there before I made the change. >>>> seems like a phatphingers error: >>>> >>>> should be: >>>> >>>> score BAYES_50 3.6 >>>> >>>> and NOT: >>>> >>>> score BAYES_50_BODY 3.6 >>> Boy, that phingers guy is a real PITA. More like phathead, as that's what >> I intended. But right after I posted I re-examined and said to myself "self >> . >> . ." >>> Thanks. Might the odd line be a result of that? >> What odd line? report looks normal > I was concerned about this: > > [score: 0.4968] > > On a line by itself. But I see similar in all headers, now that I bother to > look. > > Excuse the morning fog, please. > > joe a. >
BAYES_50 means: "Can's say if this is SPAM or HAM". I have the score
near for it. For HAM I use negative scores, for SPAM positive. But
BAYES_50 is not SPAM, nor HAM. 3.6 is way too much spammy score for it,
IMHO.
--
There is an old time toast which is golden for its beauty.
"When you ascend the hill of prosperity may you not meet a friend."
-- Mark Twain
signature.asc
Description: OpenPGP digital signature
