Twice in the last two months my bayes database has become
ineffective
(everything's marked as 99% probability of being spam).
We've got
15,000 email accounts. I'm wondering if anyone with a
similar userbase
is having the same sort of issue? I'm training hourly, no
auto-learning
(I turned that off the last time I reset bayes.) At this
point I am
forced to suspect training inaccuracy or a bad case of bayes
poisoning.
At this point I'm short on ideas, aside from doing something
drastic,
(like limiting learning to a small portion of my userbase,
cease using
bayes completely, or something else that would more or less
take a bite
out of maia's purpose), what steps might I take to
troubleshoot this issue?
Thanks for your suggestions,
Jacob Leaver
IT Manager
ReachONE Internet
_______________________________________________
Maia-users mailing list
Maia-users renaissoft.com
http://www.renaissoft.com/mailman/listinfo/maia-users
|
