Hi,

I'm trying RegEx in subject and the same text is rejected in body and in the subject even if I add a specific line for the subject

Example

Yes ... it does work.  As an example, The Expression:

((?i)Subject:(([\s]|[\!-\xB4]){0,10}[\|]){2})

will catch 2 or more "Pipes" in the subject but not in the body.  The {0,10} limits the search so that I don't get any |Loop Stack" errors.

Dan S.

Thank you Dan,

I didn't see any sample of subject filters...

I see, the ( is before the subject prefix word. That wasn't clear for me!!

It works fine!

Gaby

Gaby,

ALL Regular Expressions must be inclosed in "()" or they are not seen as RegEx's by the software but rather as a "literal".  I am glad the example helped.

Regards,

Dan S.

Dan,

Are you saying that in a keyword filter line beginning with the word Subject: you can not mix the normal wildcard matching with RegEx matching, using ( )'s?

I haven't done much with this, but I get these 2 expressions in the keywords file to do the same thing:

Subject:(cialis)

((?i)Subject:.*cialis.*)

The result being that an email with the word cialis anywhere in the subject line is blocked as spam.

Roberto and Dan could you please send me your latest, greatest keyword files also, I would greatly appreciate it.  I'm just trying to get this product to work, and from what I can see, the Bayesian Filter will not block any emails if you have a blank (or no) keyword or ip or attachments blacklist file.  My Bayesian Filter did not seem to learn anything with an "out of the box" install (no blacklist filters active, other than MAPS); all the probabilities seemed the same (400000005960464), and all the word tokens were marked good.  I thought I could just install the package, configure a few ip addresses for my servers and add my domain, and sit back and watch the Bayesian Filter learn and wipe out all my spam. 

Am I wrong in saying that a user must actively put one of the blacklist filters into action in order to get the Bayesian Filter to start learning?  And the more extensive, or complete, the keyword list is, the faster the Bayesian Filter learns?  Does it learn from MAPS rejections alone (or at all)?

If I'm right here, could you please put "you must configure a blacklist filter" into the release notes or instructions on how to set up this whole package.  Providing a good default keywords file would help, but I realize you might not want to publish it to the whole world.  This package works brilliantly, but I was certainly confused (still am?) as to why the Bayesian Filter was not marking all these "clearly spam" tokens as bad in a corpus dump.

Thanks much,

Bob

simple question...

I notice Cialis also stops "speCIALIST"  :)  if I only wanted to catch cialis would I enclose it in (cialis) ??

Lee

I answered this in another post but ...

Desperado wrote: You can use a RegEx as: ((?i)(\bcialis\b)) The \b specifies a "word Boundary" Dan S.

Dan, I tried registering on your website, but never received the authorization e-mail, because your sender domain has no mx... :-(

What website are you referring to?

Dan

mikek,

This is not "My" website.  I just attempt to maintain the users.  I do not know if Sean (the owner) can do anything about the MX as it is using dyndns service.

Dan