Byteplant Forum

Home » CleanMail Support » CleanMail SpamAssassin Filter » Teaching SpamAssassin
Teaching SpamAssassin [message #630] Tue, 17 August 2004 12:20
Mark Haller
Messages: 27
Registered: August 2004
Junior Member

Hi there,

A question which is really bugging me at present and am posting here in the hope that someone else has found an answer as I continue to scour the net ...


I want to teach SpamAssassin of ham eMails that were incorrectly flagged as SPAM.

All tools/utilities to do this point at a mail folder.

I want to point a text file at the sa-learn, as I am using Lotus Domino and can easily spit out a text file (or continuous text string) to the sa-learn.

Has anyone come across this?

Regards

Mark Haller
LogicSpot
UK
Re: Teaching SpamAssassin [message #631 is a reply to message #630] Mon, 23 August 2004 15:35 Go to previous message
support
Messages: 919
Registered: April 2004
Senior Member
> A question which is really bugging me at present and am
> posting here in the hope that someone else has found an answer
> as I continue to scour the net ...
>
>
> I want to teach SpamAssassin of ham eMails that were
> incorrectly flagged as SPAM.
>
> All tools/utilities to do this point at a mail folder.
>
> I want to point a text file at the sa-learn, as I am using
> Lotus Domino and can easily spit out a text file (or continuous
> text string) to the sa-learn.

Teaching a single message as spam:

cd [NoSpamTodayInstallationPath]
sa\sa-learn -C sa\ruleset --spam

with containing the message you want to teach.



Customer Support
Byteplant GmbH
Re: Teaching SpamAssassin [message #632 is a reply to message #631] Mon, 23 August 2004 15:54 Go to previous message
Mark Haller
Messages: 27
Registered: August 2004
Junior Member

Hey there,

Thanks for the reply - sounds great so far .... but you said with MessageFile containing the message ... do you have a sample format of such a file? or can you point me to a URL that I can learn more about this?

Sorry to bother you - I am not sure how your support extends to sa learn and so on, or whether you guys only support your core tool?

Regards

Mark Haller
LogicSpot
Re: Teaching SpamAssassin [message #633 is a reply to message #632] Fri, 27 August 2004 07:49 Go to previous message
Mark Haller
Messages: 27
Registered: August 2004
Junior Member

Hi guys,

I know you're busy but I could really use a reply to the above message .... crack this baby and we're purchasing! Just got to work out license type then! So many mailboxes in use!

Mark
Re: Teaching SpamAssassin [message #634 is a reply to message #633] Fri, 27 August 2004 10:49 Go to previous message
support
Messages: 919
Registered: April 2004
Senior Member
> I know you're busy but I could really use a reply to the
> above message .... crack this baby and we're purchasing! Just
> got to work out license type then! So many mailboxes in use!

The messages have to be in the common RFC 2822 format, see
ftp://ftp.rfc-editor.org/in-notes/rfc2822.txt
for details.



Customer Support
Byteplant GmbH
Re: Teaching SpamAssassin [message #635 is a reply to message #632] Mon, 13 September 2004 12:04 Go to previous message
InforMed Direct
Messages: 59
Registered: May 2004
Member
> with MessageFile containing the message ... do you have a

The message file is the plain internet ASCII version of the email. But as you have found, it's often darn hard to get this format of file out of a mail system. In the Outlook/Exchange world, there isn't a simple "save-as erl" function. We have to go around the houses by asking uses to move (not forward!) spam emails into a shared folder. Then once a day, an admin person downloads these into Outlook Express and then we use a free utility DBxtract which can export an Outlook Express inbox back into the right format for feeding through sa-learn.

The same process could be used for ham although we find that as NST/SA automatically passes non-spam email through the learner, we don't have to worry about ham too much.

I think that the latest version (paid upgrade) of NST has a mechanism whereby users can forward spam email to a special address which NST then automatically passes through the sa-learn process as spam. We've not got this version so can't comment more.

Cheers, Rob.

Re: Teaching SpamAssassin [message #636 is a reply to message #635] Mon, 13 September 2004 12:13 Go to previous message
Mark Haller
Messages: 27
Registered: August 2004
Junior Member

Hi Rob,

Wow, sounds like a convoluted way that you have to manually intervene. Real shame Sad Thinking on your feet there!

Can someone from Byteplant confirm Rob's final point - the new version has this functionality available - as if that's the case, the developer working on this can ditch his work! He's having real problems trying to extract a valid format as well - simple mails are easy but more complex ones that have come in, in different formats, are proving tough. We've just bought the full version after careful thought so hope this in!


Cheers,

Mark
Re: Teaching SpamAssassin [message #637 is a reply to message #636] Mon, 13 September 2004 12:16 Go to previous message
InforMed Direct
Messages: 59
Registered: May 2004
Member
> Can someone from Byteplant confirm Rob's final point - the
> new version has this functionality available - as if that's the
> case, the developer working on this can ditch his work! He's

I infer it's the honeypot mechanism mentioned on the release notes:

http://www.byteplant.com/company/news/04081901.html

Cheers, Rob.

Re: Teaching SpamAssassin [message #638 is a reply to message #637] Mon, 13 September 2004 12:19 Go to previous message
Mark Haller
Messages: 27
Registered: August 2004
Junior Member

Hey Rob,

You know what - you'll be right there, then!! Smile

I was thinking Winnie the Pooh and the Te of Piglet (great book by the way) among others and didn't even realise!

The beta testing starts a week on Monday so we'll see how well it fairs! Smile


Thanks again,

Mark
LogicSpot
Re: Teaching SpamAssassin [message #639 is a reply to message #638] Mon, 04 October 2004 17:22 Go to previous message
support
Messages: 919
Registered: April 2004
Senior Member
Now, to finally say something on this subject:

Learning forwarded mails is at best useless, at worst even counter-productive:
Upon forwarding, both mail clients (and mail servers) modify mails in unforeseeable ways. Both sa-learn or NoSpamToday! are unable to undo the changes your mail client, and the forwarding option you've chosen, may have introduced.

With version 2 we intend to introduce a mail storage. This mail storage can be used to cache incoming mails, in a pristine and unmodified condition, and in a suitable format for learning.

To learn messages, the mail administrator would collect storage IDs of messages to learn as either ham or spam, and use those IDs to sa-learn the offending mails from the cache.

This is not yet the whole way, but at least you no longer have to trouble yourself with getting mails exported and converted into MIME format from proprietary database formats.



Customer Support
Byteplant GmbH
Re: Teaching SpamAssassin [message #640 is a reply to message #639] Tue, 05 October 2004 17:54 Go to previous message
InforMed Direct
Messages: 59
Registered: May 2004
Member
> Learning forwarded mails is at best useless, at worst even
> counter-productive:
> Upon forwarding, both mail clients (and mail servers) modify
> mails in unforeseeable ways. Both sa-learn or NoSpamToday! are

That's why we don't forward spam, we *move* it.

Cheers, Rob

Previous Topic: Whitelist - how?
Next Topic: cant get sa-learn to work, what am i doing wrong
Goto Forum:
  


Current Time: Sun Dec 04 19:22:20 CET 2016