spamassassin doesn't seem to be using bayes

[jdow]

From: <akonstam at trinity.edu>

> On Fri, Oct 21, 2005 at 05:15:56PM +0200, Alexander Dalloz wrote:

>> Did you set in local.cf something like following?
>> 
>> use_bayes 1
>> auto_learn 1
>> bayes_learn_to_journal 1
>> bayes_file_mode 0600
>> bayes_path /var/spool/MIMEDefang/bayes
>> 
>> > Best, Darren
>> 
> 
> The lines like the ones above appear in:
> .spamassassin/user_prefs assuming these are personal user
> settings.

I place those lines in the local.cf file. These are the only two I
keep in user_prefs.
---8<---
# How many hits before a mail is considered spam.
required_hits           5.0
rewrite_header Subject     *****SPAM***** _SCORE(00)_ **
---8<---

The first is obvious. The second gives me spam headers that start
with things like "*****SPAM***** 038.7 **". This makes sorting spam
by score easy so I can double check the low scores in case some ham
gets mismarked.

> In my instructions auto_learn appears as bayes_auto_learn. The
> last three lines have default values so I would assume they are
> not needed.
> 
> One more thing the option: bayes_auto_learn_threshold_spam needs to be
> considered since a spam score below this threshold is
> not auto-learned to be spam. Its default in 12.0.

I turn off auto-learn altogether. A mislearned spam or ham is a pain
in the donkey to repair. At least set the thresholds MUCH wider,
especially if you visit http://www.rulesemporium.com/ and install a
large selection of rules. (I run some 42 sets of rules. That MAY be
a little overboard. But at present aside from some LKML mailings
about one in 1000 hams is mismarked and about one in 500 spams gets
through unmarked. These are usually new spam formats for spam with
you as one of the first recipients. The BL test rules generally lead
to even new spam formats being marked as spam.)

> Similarly there is a:bayes_auto_learn_threshold_nonspam. Default
> is 0.1

{^_^}   Joanne