Are the Bayesian learning tools sa-learn-spam, sa-learn-nonspam, and
sa-rebuild in working order?
I want to set up Bayesian for a whole mailhost, rather than for a single
user. I want to put the databases in a location under /var/spool, not in a
user's home directory.
So I have my nonspam and spam blobs of mail to learn on. I prepared them
from my collection of spam stored on my Cyrus IMAP server, and my regular
current INBOX and some folders containing mailing list messages. There are
15491 messages (63 Mb) in the nonspam corpus, and 13197 messages (61 Mb) in
the spam corpus. I just concatenated them all together into one file for
each corpus. Cyrus stores mail in RFC822 format, no From_ header as with
mbox format. I grepped out any previous MIMEDefang or SpamAssassin headers
before processing. I created a directory /var/spool/SA_Bayes and
configured spamassassin.cf with "bayes_path /var/spool/SA_Bayes/bayes".
Then I ran "sa-learn-spam -D --showdots --norebuild
--config-file=/etc/mail/spamassassin.cf --file spam.corpus"
The first time I ran this command, it created ~/.spamassassin/ and put some
files bayes_toks and bayes_seen in there. It didn't create
/var/spool/SA_Bayes/bayes_toks as expected. So I killed it, and tried
again after su - root (instead of just su root). This time it created
/.spamassassin/ but did not create any bayes_* files there. It also did
not seem to be doing anything. I tried it again as my regular user, and
ther same thing happened - it created ~/.spamassassin/ but did not put any
files there.
So I am now wondering whether I should expect any of this to work yet in
the unreleased version 2.50. How far along is the Bayseian code? Is it
useable? What should I expect from it? How is the learning process
supposed to go, and will I just have to move the bayes_ files to my
/var/spool location to get SA to use them in the location specified in
spamassassin.cf?
---
"The avalanche has already begun. It is too late for the pebbles to vote."
-- Kosh
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
