[Spambayes] Unique text = ham?
Thomas Hruska
thruska at cubiclesoft.com
Mon Mar 27 03:50:17 CEST 2006
Amedee Van Gasse wrote:
> Thomas Hruska wrote:
>> I need a way to tell SpamBayes to ignore classifying e-mails that
>> contain a specific sequence of characters even if it would have
>> classified the e-mail as spam. How do I go about accomplishing this
>> task with the POP3 Proxy SpamBayes server? I can't seem to locate an
>> option to do this - perhaps I missed it? This feature is a "stupid"
>> feature, but will keep the SpamBayes database lean-and-mean and a fair
>> balance between ham and spam for training. A 20:1 ham to spam ratio
>> instead of 1000:1. I can't have it classifying these critical e-mails
>> at all. I want them all even if they ARE spam.
>
> I think, if you know the mathematical theory behind spambayes, if it is
> a unique sequence of characters, and that unique sequence of characters
> will never appear in spam, it's just the law of large numbers that applies.
> So, if you keep on telling that this specific kind of email is ham,
> after just a few mails sb will learn. Stupid large numbers always beat
> smart and stupid features.
>
> But I don't know how the pop3 proxy implementation works.
> At work I use the Oulook plugin, and at home I use the procmail filter.
> (I mainly use Linux and try to avoid Windows if I can)
>
Except I want ANY message that uses the sequence of characters,
_INCLUDING_ spam. The unique sequence of characters _WILL_ appear in
spam, but I still want those messages anyway. I can't have even the
_slightest_ possibility that SpamBayes will flag those messages as spam.
I realize that wanting spam with an anti-spam tool seems
counter-intuitive, but I have my reasons for needing this.
--
Thomas Hruska
CubicleSoft President
Ph: 517-803-4197
Safe C++ Design Principles (First Edition)
Learn how to write memory leak-free, secure,
portable, and user-friendly software.
Learn more and view a sample chapter:
http://www.CubicleSoft.com/SafeCPPDesign/
More information about the SpamBayes
mailing list