Can I bump up the scores?

[Pat Traynor <pa...@ssih.com>]

A lot of spam is slipping through.  I was wondering how dangerous it
would be for me to bump up some of these scores.  This is from a typical
spam I've been getting:

    X-Spam-Level: ***
    X-Spam-Status: No, hits=3.3 required=5.0
tests=BAYES_50,HTML_MESSAGE,
        RCVD_IN_BL_SPAMCOP_NET,RCVD_IN_DSBL,RCVD_IN_SORBS autolearn=no 
        version=2.60-rc3

Looking at the scores for some of these, I see:

    score RCVD_IN_BL_SPAMCOP_NET 0 2.95 0 2.00
    score RCVD_IN_DSBL 0 1.501 0 1.106

Perhaps this is an invalid assumption, but I'm assuming that since
they've calculated the score to a one-thousandth of a point, there is
some good reason for that particular score, and maybe I shouldn't be
monkeying with it.

But I'd love to go in and just bump up some of these scores a solid
point or two.  Am I asking for trouble by doing this?

On a related subject, are there updated scorecards that some of you
wizards have tweaked that could be downloaded?

Thanks for any help.

--pat--
-- 
Pat Traynor
pat@ssih.com

Re: Can I bump up the scores?

[Bob George <ma...@ttlexceeded.com>]

Pat Traynor wrote:
> [...]
> Somehow I thought that bayes was a "single user" thing.  i.e.,
if I
> do a great job training bayes, that'll help me fine.  But it
does
> nothing for the other people that receive mail on my machine.
Am I
> wrong?
You *can* set up site-wide bayes, or per-user. The trick with
site-wide is that one person's spam is another's critical daily
fix. Just as with tuning the SA rule set, you can have it affect
all users, or just a single user.

Depending on the spam that's getting through, you might try some
of the additional rule sets mentioned frequently on this list
(bigevil, backhair, weeds, etc.) They'll probably give it the
'nudge' you need without causing too many false positives.

Let me emphasize that I'm not a particular expert on this topic,
just someone who's got SA tuned to handle the situation you
described with minimal fuss!

- Bob

Re: Can I bump up the scores?

[Pat Traynor <pa...@ssih.com>]

On Mon, 2 Feb 2004, Bob George wrote:

> Pat Traynor wrote:
> > A lot of spam is slipping through.  I was wondering how
> dangerous it
> > would be for me to bump up some of these scores.  This is from a
> > typical spam I've been getting:
> >
> >     X-Spam-Level: ***
> >     X-Spam-Status: No, hits=3.3 required=5.0
> > tests=BAYES_50 [...]
> 
> Are you training bayes with those that DO slip through? The "rules
> du jour" set helped set things right as well (see archives) by
> detecting some of the sneakier spam. This combination worked very
> well for me, with no need to tweak any default scores.

Somehow I thought that bayes was a "single user" thing.  i.e., if I do a
great job training bayes, that'll help me fine.  But it does nothing for
the other people that receive mail on my machine.  Am I wrong?

--pat--
-- 
Pat Traynor
pat@ssih.com

Re: Can I bump up the scores?

[Bob George <ma...@ttlexceeded.com>]

Pat Traynor wrote:
> A lot of spam is slipping through.  I was wondering how
dangerous it
> would be for me to bump up some of these scores.  This is from a
> typical spam I've been getting:
>
>     X-Spam-Level: ***
>     X-Spam-Status: No, hits=3.3 required=5.0
> tests=BAYES_50 [...]

Are you training bayes with those that DO slip through? The "rules
du jour" set helped set things right as well (see archives) by
detecting some of the sneakier spam. This combination worked very
well for me, with no need to tweak any default scores.

- Bob

Re: Can I bump up the scores?

[Andy Spiegl <sp...@spiegl.de>]

Hi Pat,

> I believe that you aren't using the bayes filter.
Sorry, just saw that I was wrong!  You already _are_ using Bayes:
> tests=BAYES_50 [...]

Sorry for the noise,
 Andy.

-- 
                              o      _     _         _
  ------- __o       __o      /\_   _ \\o  (_)\__/o  (_)          -o)
  ----- _`\<,_    _`\<,_    _>(_) (_)/<_    \_| \   _|/' \/       /\\
  ---- (_)/ (_)  (_)/ (_)  (_)        (_)   (_)    (_)'  _\o_    _\_v
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Great spirits have always encountered violent opposition from
 mediocre minds.   - Albert Einstein

Re: Can I bump up the scores?

[Andy Spiegl <sp...@spiegl.de>]

Hi Pat,

sorry, can't help you with your question but as a side note:
I believe that you aren't using the bayes filter.  Try it out, I have
really great results with it.
 Andy.

-- 
                              o      _     _         _
  ------- __o       __o      /\_   _ \\o  (_)\__/o  (_)          -o)
  ----- _`\<,_    _`\<,_    _>(_) (_)/<_    \_| \   _|/' \/       /\\
  ---- (_)/ (_)  (_)/ (_)  (_)        (_)   (_)    (_)'  _\o_    _\_v
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 A bicycle can't stand on its own because it is two-tired.