On Dec 3, 2003, at 11:18 AM, Ernest W. Lessenger wrote:
Are all of your users in the same company/organization? If so, you
could use the same database for all of them for a while, and then
split it out once it kicks in. It's still a better idea to train
individually, but your users (may) have mo
I'm setting up per user bayesian databases but as i've mentioned in
other posts, many of the people using the system recieve low volumes of
spam... 4-10 a day maybe. Accordingly it would take them 1-2 months for
the database to kick in.
Does anybody have a bayesian database that is fairly gener
I'm getting a common response here which is to lean towards bayes
rather than score manipulation.
However, bayes takes so long to build up. Especially if grandma only
gets 2 or 3 spams a day. Given that SA refuses to use bayesian
filtering til after it has collected 200 spam samples, in my exam
Hi Everybody,
This is a long post but i would REALLY appreciate your input,
criticism, flaming, and what not.
I'm working on integrating spamassassin into our own spam filtering
mechanism. Currently, with a score of 5 or greater we modify the
subject line to indicate the spammishness of the me
faith in them? And is a 9
a little late? Sure if 100k people have seen it we can likely consider
it spam, but why 100k? why not less?
-Arlo
On Wednesday, October 22, 2003, at 01:35 AM, Larry Gilson wrote:
Hi Arlo,
-Original Message-
From: Arlo Gilbert [mailto:[EMAIL PROTECTED]
Sent
I'm hopeful that somebody can explain the dcc results to me.
i understand the concept of dcc, however spamassassin docs say that
you should set limits that are numeric. eg. so that dcc doesnt
give false positives.
upon looking at the headers however, i see the following:
X-Spam-DCC: SINE
it would appear from the data im seeing that bayes is learning the to
and received headers on mails... obviously this seems a bit redundant
and will only add to the size of the bayes db, without contributing
anything (maybe even harming the learning?) of the bayes engine.
any thoughts on this?
Well no thanks to any of the documentation or this list for that
matter, but it appears that the implementation of SA's
mail::spamassassin module is not all it says in the docs when using
mysql based user prefs.
Spamassassin.pm has a non public sub called init that is called
internally from th
does a "required_hits" directive in the local.cf override any user
prefs that would be found in the mysql user prefs database?
i've confirmed that mail::spamassassin is retrieving the pref for the
correct user, but without fail spamassassin uses the system wide
required_hits.
i'm at the end of
this is kind of a repost. sorry, my last post of it wasnt very clear.
perl 5.8, sa 2.6, redhat9
caling the following: my $spamtest =
Mail::SpamAssassin->new(dont_copy_prefs => '1')
gives the following error
Can't bless non-reference value at
/usr/lib/perl5/site_perl/5.8.0/Mail/SpamAssassin.pm
Spamassassin 2.6, perl 5.8 on a redhat 9 machine
called from perl i get the following debug messages even though it
appears to read those two settings as sa does abide by them.
debug: Failed to parse line in SpamAssassin configuration, skipping:
report_header 1
debug: Failed to parse line in Sp
Running it from perl with debug enabled...Settings all properly
configured in the /etc/mail/spamassassin/local.cf and it wont read them.
Attempting to pass the info into the new mail::spamassassin via (
user_scores_dsn => 'DBI:whatever:whatever:whatever' ) gives me a bunch
of bless errors and b
12 matches
Mail list logo
