You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by David Baron <d_...@012.net.il> on 2008/01/30 07:38:08 UTC

Re: PUPPIES FOR SALE

On Wednesday 30 January 2008, debian-user-digest-request@lists.debian.org 
wrote:
> > First of all, let me say a big thank you, as it proves that my spamfilter
> > is not yet filtering out spam from the Debian mailing list.
>
> I'd be interested in a filter that would work on a message like that.. Not
> sure what the difference is between a message about puppies and a message
> about ndiswrapper.. Not sure if ndiswrapper is in the dictionary, but
> puppies is. If someone DIDN'T get that last message, would you please raise
> your hand??

OK spamassassin folks: Rules which would say no puppies on software mailing 
lists, no software on dog-breeders mailing lists. A few false alarms, i.e. 
"that great new app is such a sweet-puppie" and that "breeder's management 
package is a killer app" (or is that a yap?).



Re: PUPPIES FOR SALE

Posted by mouss <mo...@netoyen.net>.
John Hardin wrote:
> On Wed, 2008-01-30 at 08:38 +0200, David Baron wrote:
>
>   
>> OK spamassassin folks: Rules which would say no puppies on software mailing 
>> lists, no software on dog-breeders mailing lists. A few false alarms, i.e. 
>> "that great new app is such a sweet-puppie" and that "breeder's management 
>> package is a killer app" (or is that a yap?).
>>     
>
> (1) Some __ rules to detect the mailing lists from the headers (assuming
> the list manager puts in nice mailing list headers), like
>    header __LIST_DEBIAN List_ID =~ /\.debian\./
>
> (2) Some content-specific __ rules, like 
>    body __PUPPIES /\bpupp(?:y|ies)\b/i
>
> (3) meta them together for scoring
>    meta DEBIAN_PUPPIES (__LIST_DEBIAN && __PUPPIES)
>    score DEBIAN_PUPPIES 1.00
>
> Repeat as needed.
>   

so we can no more discuss Puppy Linux or the Puppy package manager on 
debian lists?

keyword filtering on general public lists is risky. I wonder if training 
bayes with a large corpus would help (the problem is what spam to use in 
the corpus).



Re: PUPPIES FOR SALE

Posted by John Hardin <jh...@impsec.org>.
On Wed, 2008-01-30 at 08:38 +0200, David Baron wrote:

> OK spamassassin folks: Rules which would say no puppies on software mailing 
> lists, no software on dog-breeders mailing lists. A few false alarms, i.e. 
> "that great new app is such a sweet-puppie" and that "breeder's management 
> package is a killer app" (or is that a yap?).

(1) Some __ rules to detect the mailing lists from the headers (assuming
the list manager puts in nice mailing list headers), like
   header __LIST_DEBIAN List_ID =~ /\.debian\./

(2) Some content-specific __ rules, like 
   body __PUPPIES /\bpupp(?:y|ies)\b/i

(3) meta them together for scoring
   meta DEBIAN_PUPPIES (__LIST_DEBIAN && __PUPPIES)
   score DEBIAN_PUPPIES 1.00

Repeat as needed.

--
 John Hardin KA7OHZ                    http://www.impsec.org/~jhardin/
 jhardin@impsec.org    FALaholic #11174     pgpk -a jhardin@impsec.org
 key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C  AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
  Efficiency can magnify good, but it magnifies evil just as well.
  So, we should not be surprised to find that modern electronic
  communication magnifies stupidity as *efficiently* as it magnifies
  intelligence.                                   -- Robert A. Matern
-----------------------------------------------------------------------
 2 days until the 5th anniversary of the loss of STS-107 Columbia