You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Pieter Vanmeerbeek <pi...@able.be> on 2006/04/18 09:47:33 UTC

Bayesian learning and corrective training

Hi,

I've got a question on Bayesian learning using the script explained here:
http://www.jousset.org/pub/sa-postfix.en.html

The plan is to let users drag multiple spam messages in a new message and
sent this to a dedicated e-mail address. On the website mentioned earlier it
is said that you need a message/rfc822 content type. I checked this using
Microsoft outlook and it seems to be correct.

Can anyone confirm that this will always be the case when dragging multiple
messages into a new one? And will this be so in both outlook and outlook
express?

Kind regards,
Pieter


--
---------------------------------------------------
aXs GUARD Training Center 
more info at http://www.axsguard.com/indextraining.htm

aXs GUARD has completed security and anti-virus checks on this e-mail (http://www.axsguard.com)
---------------------------------------------------
Able NV: ond.nr 0457.938.087
RPR Mechelen


RE: Bayesian learning and corrective training

Posted by Herb Martin <He...@learnquick.com>.
> gedoesn't outlook/exchange remove/rewrite some headers at reception 
> time, thus making it impossible to retriev the original message?

This method has worked well for us but due
to your specific question I diff'd some originals
against the forwarded copies.

There are SOME differences -- it seems Some received headers,
and oddly the SpamAssassin score report (probably due to length)
are dropped, some re-ordering is done.

Generally the headers are there (unless one needs something
peculiar to those received headers.)


--
Herb Martin


Re: Bayesian learning and corrective training

Posted by mouss <us...@free.fr>.
Herb Martin wrote:
>>> Can anyone confirm that this will always be the case when dragging
>>>       
>> multiple
>>     
>>> messages into a new one? And will this be so in both 
>>>       
>> outlook and outlook
>>     
>>> express?
>>>       
>> I can't confirm that it WILL always be true, but I believe 
>> that it mostly
>> *should* always be true.  I've known Outlook to send attached 
>> "*.msg" files
>> instead of a properly encoded attachment.  I'm not positive under the
>> conditions that this would occur.  I don't think I've ever 
>> seen OE do this,
>> but I suppose it probably could.
>>
>> I think as long as you ignore improper attachments you should 
>> be in pretty
>> good shape.
>>     
>
> The current version of Outlook 2003 SP1, and likely
> others follow (at least these rules:
>
> If you select MULTIPLE messages and hit Forward then
> the messages get properly (with full headers as received)
> attached -- BUT if you select one message the message 
> gets 'edited' as the text of the outgoing message.
>
> If you drag a message into a draft message it SEEMS
> to get properly attached as above.
>
> My method is to ALWAYS select multiple messages (even
> when only one is desired) and then remove any 'extras'
> that I do not wish to send.
>
> A simple perl script (just a couple of lines) will split 
> the message(s) back into individual messages that can
> be properly used for feeding Bayes or other classifiers.
>   
gedoesn't outlook/exchange remove/rewrite some headers at reception 
time, thus making it impossible to retriev the original message?

RE: Bayesian learning and corrective training

Posted by Herb Martin <He...@learnquick.com>.
> > Can anyone confirm that this will always be the case when dragging
> multiple
> > messages into a new one? And will this be so in both 
> outlook and outlook
> > express?
> 
> I can't confirm that it WILL always be true, but I believe 
> that it mostly
> *should* always be true.  I've known Outlook to send attached 
> "*.msg" files
> instead of a properly encoded attachment.  I'm not positive under the
> conditions that this would occur.  I don't think I've ever 
> seen OE do this,
> but I suppose it probably could.
> 
> I think as long as you ignore improper attachments you should 
> be in pretty
> good shape.

The current version of Outlook 2003 SP1, and likely
others follow (at least these rules:

If you select MULTIPLE messages and hit Forward then
the messages get properly (with full headers as received)
attached -- BUT if you select one message the message 
gets 'edited' as the text of the outgoing message.

If you drag a message into a draft message it SEEMS
to get properly attached as above.

My method is to ALWAYS select multiple messages (even
when only one is desired) and then remove any 'extras'
that I do not wish to send.

A simple perl script (just a couple of lines) will split 
the message(s) back into individual messages that can
be properly used for feeding Bayes or other classifiers.


Herb Martin, MCSE, MVP
HerbM@LearnQuick.Com http://LearnQuick.Com
512 388 7339   -or-   1 800 MCSE PRO
Accelerated MCSE in a Week Seminars 

> -----Original Message-----
> From: Loren Wilton [mailto:lwilton@earthlink.net] 
> Sent: Tuesday, April 18, 2006 2:56 AM
> To: users@spamassassin.apache.org
> Subject: Re: Bayesian learning and corrective training
> 
> 
>         Loren
> 


Re: Bayesian learning and corrective training

Posted by Loren Wilton <lw...@earthlink.net>.
> Can anyone confirm that this will always be the case when dragging
multiple
> messages into a new one? And will this be so in both outlook and outlook
> express?

I can't confirm that it WILL always be true, but I believe that it mostly
*should* always be true.  I've known Outlook to send attached "*.msg" files
instead of a properly encoded attachment.  I'm not positive under the
conditions that this would occur.  I don't think I've ever seen OE do this,
but I suppose it probably could.

I think as long as you ignore improper attachments you should be in pretty
good shape.

        Loren