I made the mistake of interrupting an expiry run with a ctrl-C. This
had the effect of zeroing out my bayes_msgcount file, and check_bayes_db
now shows:
0.000 0 74 0 non-token data: nspam
0.000 0 5 0 non-token data: nham
0.000 0 50315 0 non-token data: ntokens
0.000 0 0 0 non-token data: oldest age
0.000 0 139 0 non-token data: current scan-count
I'd trained it on about 3500 messages, mixed spam and ham. The tokens
themselves are fine -- it's just the message count that's broken (and
currently below the threshold for activating the Bayesian filter).
Moreover, because I have fed my current mail archive to SpamAssassin
in the past, it cannot re-learn from those messages -- for the vast
majority of the messages, it concludes it has nothing to learn, and
for the remainder, I'm concerned that I'm reinforcing the existing
SpamAssassin Bayes probabilities.
How do I fix this? I don't know exactly how many messages, spam, or ham,
there have been, but if I could put in a reasonable estimate, that would
be fine. Alternatively, forcing Bayes to turn back on would do, though
possibly less desirable.
-- Lydia
-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
