Re: spam filter problem

2002-05-21 Thread James R. Van Zandt


Karl Runge writes:
It is possible you have an oddly named file   ^Subject: (ADV:)
that contains those missing ~ 100 emails of yours.

You're right!

  $ ls -l *Sub*
  -rw---1 jrv  jrv   2555170 May 13 19:26 ^Subject:

Thank you very much!

  - Jim Van Zandt


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-20 Thread James R. Van Zandt


Thanks for all the replies.

need a * here, don't you?
 ^Subject: (ADV:)

Right!  Ben Boulanger and Karl Runge spotted this one.

I rewrote it this way:

:0 H # recognize junk mail by subject
* ^Subject: (ADV:)
Mail/junk-subject

With this change, the rules seem to be working.

 :0 H # foreign language junk mail
 * charset=.ks_c_5601-1987.

Not sure about this one... what's the actual header line look like
for this?  \. is needed if you want to match a ., but.. I'm not
familiar with the header you're going for..

Maybe a quarter of my spam is in some foreign language, and includes
a header line like this:

  Content-Type: text/html; charset=ks_c_5601-1987

I've also added this rule

:0 B # some worm
* TVqQAAME//8AALgAQAAA
Mail/junk-worm

...which caught four messages out of the last batch.  But that was
over half a megabyte of data!  I've since redirected it to /dev/null.

Thanks again!

  - Jim Van Zandt


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-20 Thread James R. Van Zandt


Rich Payne [EMAIL PROTECTED] writes:

 you may want to have a look at razor
 (http://razor.sourceforge.net/). It's a distributed SPAM checking
 system.

I noticed this a while back, and it looks very interesting.  However
just the other day I read a comment at Slashdot that someone has been
poisoning the razor database, so that it labeled some legitimate
mailing list postings as spam.  

 There's also a razor-report program that allows you to feed new SPAM 
 messages into the systems. 

I've been hoping someone would integrated this into RMAIL in Emacs, so
with a single keystroke I could delete the mail and report it as spam
at the same time.  Of course, that also makes it easy to report by
mistake.  Maybe that's what's behind the Slashdot comment.  I suppose
one could require two spam reports or something.

Mark Komarinski [EMAIL PROTECTED] writes:

 I heartily recommend spamassassin.  It used a variety of weightings to
 see if the mail you have is spam.  For example, if the mail is listed
 in Razor, it's worth 2 points...

Sounds like the next step.  Thanks for the pointers!

- Jim Van Zandt


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-20 Thread Rich Payne

  (http://razor.sourceforge.net/). It's a distributed SPAM checking
  system.
 
 I noticed this a while back, and it looks very interesting.  However
 just the other day I read a comment at Slashdot that someone has been
 poisoning the razor database, so that it labeled some legitimate
 mailing list postings as spam.  

Hmmm..while I don't know the whole razor system, so far I've not had any 
legit emails put into my SPAM folder. I don't delete the SPAM but rather 
redirect it and go through the folder every few days. I think it probably 
requires x reports from y different sites before a message gets added.

As for slashdot, I beleive things posted on there about as much as I 
beleive the weekly world news (entertaining at times yes, but factual, I 
doubt it).

--rdp

-- 
Rich Payne
http://talisman.mv.com


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



spam filter problem

2002-05-17 Thread James R. Van Zandt


I've been running a simple procmail filter to get rid of spam from
some specific sites.  The sample below only includes a few of the
addresses, but even with the whole list it's no longer very effective.
Last weekend I decided to tune it up to filter out more of the spam.
I added the last three rules below.  Out of the next batch of 116
emails, these rules eliminated 116.  That's right - all the babies got
thrown out with the bath water.  Of course, I've commented out the new
rules until I debug them.

If someone has a non-risky way to test procmail rules, I'd appreciate
hearing about it.

Of course, if you see what I've done wrong I'd like to hear that too.

- Jim Van Zandt



:0 Whc: msgid.lock
| formail -D 65536 .msgid.cache
:0
* 
^From:.*(assetmanagement.net|[EMAIL PROTECTED]|moreorders.com|[EMAIL PROTECTED]|[EMAIL PROTECTED]|[EMAIL PROTECTED])|^Subject:.*\$\$\$|^To:
 [EMAIL PROTECTED]
|egrep '^(Subject|From|To):' Mail/junkheaders
:0 B # recognize junk mail by content
* (so dilghted when i|This game is my first work.)
/dev/null
:0 H # recognize junk mail by subject
^Subject: (ADV:)
/dev/null
:0 H # foreign language junk mail
* charset=.ks_c_5601-1987.
/dev/null


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-17 Thread Rich Payne



I have no idea what's wrong with this...however you may want to have a 
look at razor (http://razor.sourceforge.net/). It's a distributed SPAM 
checking system. Basically you don't  have to worry about keeping a list 
of the senders etcyou just use procmail to pass all your mail through 
razor-check, if it matches a known SPAM message you can have procmail do 
as you like with (direct to /dev/null etc...).

There's also a razor-report program that allows you to feed new SPAM 
messages into the systems. While it doesn't work 100%, it does catch a lot 
of the usual stuff going around.

--rdp

On Thu, 16 May 2002, James R. Van Zandt wrote:

 
 I've been running a simple procmail filter to get rid of spam from
 some specific sites.  The sample below only includes a few of the
 addresses, but even with the whole list it's no longer very effective.
 Last weekend I decided to tune it up to filter out more of the spam.
 I added the last three rules below.  Out of the next batch of 116
 emails, these rules eliminated 116.  That's right - all the babies got
 thrown out with the bath water.  Of course, I've commented out the new
 rules until I debug them.
 
 If someone has a non-risky way to test procmail rules, I'd appreciate
 hearing about it.
 
 Of course, if you see what I've done wrong I'd like to hear that too.
 
   - Jim Van Zandt
 
 
 
 :0 Whc: msgid.lock
 | formail -D 65536 .msgid.cache
 :0
 * 
^From:.*(assetmanagement.net|[EMAIL PROTECTED]|moreorders.com|[EMAIL PROTECTED]|[EMAIL PROTECTED]|[EMAIL PROTECTED])|^Subject:.*\$\$\$|^To:
 [EMAIL PROTECTED]
 |egrep '^(Subject|From|To):' Mail/junkheaders
 :0 B # recognize junk mail by content
 * (so dilghted when i|This game is my first work.)
 /dev/null
 :0 H # recognize junk mail by subject
 ^Subject: (ADV:)
 /dev/null
 :0 H # foreign language junk mail
 * charset=.ks_c_5601-1987.
 /dev/null
 
 
 *
 To unsubscribe from this list, send mail to [EMAIL PROTECTED]
 with the text 'unsubscribe gnhlug' in the message body.
 *
 

-- 
Rich Payne
http://talisman.mv.com


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-17 Thread Ben Boulanger

On Thu, 16 May 2002, James R. Van Zandt wrote:
 If someone has a non-risky way to test procmail rules, I'd appreciate
 hearing about it.

Don't send to /dev/null at first, send to something you can get to with 
your mail reader - ~/mail/filtered or something usually works for me.

 :0 H # recognize junk mail by subject
need a * here, don't you?
 ^Subject: (ADV:)

 /dev/null
 :0 H # foreign language junk mail
 * charset=.ks_c_5601-1987.

Not sure about this one... what's the actual header line look like for 
this?  \. is needed if you want to match a ., but.. I'm not familiar 
with the header you're going for..

Ben

-- 

Better a diamond with a flaw than a pebble without one. 


*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-17 Thread Mark Komarinski

I heartily recommend spamassassin.  It used a variety of weightings to
see if the mail you have is spam.  For example, if the mail is listed
in Razor, it's worth 2 points, and if it came from a site listed in one
of the RBLs, it's worth a few points, and so on.  You can configure the
weighting as you like (and the threshold where it's considered spam).

Once all that's done, a simple .procmailrc setup and you're done.

-Mark  

On Fri, 2002-05-17 at 12:06, Rich Payne wrote:
 
 
 I have no idea what's wrong with this...however you may want to have a 
 look at razor (http://razor.sourceforge.net/). It's a distributed SPAM 
 checking system. Basically you don't  have to worry about keeping a list 
 of the senders etcyou just use procmail to pass all your mail through 
 razor-check, if it matches a known SPAM message you can have procmail do 
 as you like with (direct to /dev/null etc...).
 
 There's also a razor-report program that allows you to feed new SPAM 
 messages into the systems. While it doesn't work 100%, it does catch a lot 
 of the usual stuff going around.
 
 --rdp
 
 On Thu, 16 May 2002, James R. Van Zandt wrote:
 
  
  I've been running a simple procmail filter to get rid of spam from
  some specific sites.  The sample below only includes a few of the
  addresses, but even with the whole list it's no longer very effective.
  Last weekend I decided to tune it up to filter out more of the spam.
  I added the last three rules below.  Out of the next batch of 116
  emails, these rules eliminated 116.  That's right - all the babies got
  thrown out with the bath water.  Of course, I've commented out the new
  rules until I debug them.
  
  If someone has a non-risky way to test procmail rules, I'd appreciate
  hearing about it.
  
  Of course, if you see what I've done wrong I'd like to hear that too.
  
  - Jim Van Zandt
  
  
  
  :0 Whc: msgid.lock
  | formail -D 65536 .msgid.cache
  :0
  * 
^From:.*(assetmanagement.net|[EMAIL PROTECTED]|moreorders.com|[EMAIL PROTECTED]|[EMAIL PROTECTED]|[EMAIL PROTECTED])|^Subject:.*\$\$\$|^To:
 [EMAIL PROTECTED]
  |egrep '^(Subject|From|To):' Mail/junkheaders
  :0 B # recognize junk mail by content
  * (so dilghted when i|This game is my first work.)
  /dev/null
  :0 H # recognize junk mail by subject
  ^Subject: (ADV:)
  /dev/null
  :0 H # foreign language junk mail
  * charset=.ks_c_5601-1987.
  /dev/null
  
  
  *
  To unsubscribe from this list, send mail to [EMAIL PROTECTED]
  with the text 'unsubscribe gnhlug' in the message body.
  *
  
 
 -- 
 Rich Payne
 http://talisman.mv.com
 
 
 *
 To unsubscribe from this list, send mail to [EMAIL PROTECTED]
 with the text 'unsubscribe gnhlug' in the message body.
 *



*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*



Re: spam filter problem

2002-05-17 Thread Cole Tuininga

On Fri, 2002-05-17 at 12:51, Mark Komarinski wrote:
 I heartily recommend spamassassin.  It used a variety of weightings to
 see if the mail you have is spam.  For example, if the mail is listed
 in Razor, it's worth 2 points, and if it came from a site listed in one
 of the RBLs, it's worth a few points, and so on.  You can configure the
 weighting as you like (and the threshold where it's considered spam).

I'll second this.  

I've been using spam assassin in conjunction with razor for about a
month now.  Only 1 or 2 false positives, and I'd say there's been about
97% rate of correct catches.

-- 
Some Axiom: The intelligence of the world is a constant.  
The population is increasing.

Cole Tuininga
Lead Developer
Code Energy, Inc
[EMAIL PROTECTED]
PGP Key ID: 0x43E5755D



*
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*