On Thu, 2011-03-03 at 19:40 -0500, Dennis German wrote:
> Can someone comment on the low score assigned to the email located at
>
> http://www.cccu.us/hundredThousand.txt
>
> X-Spam-testscores: AWL=1.086,BAYES_00=-2.599,HTML_MESSAGE=0.001,
> MILLION_USD=1.528
>
> Is my bayes "broken"?
I'd phrase it "badly trained" for this type of spam.
The BAYES_00 hit suggests, that similar fraud spam has been incorrectly
trained in the past. This might have been by auto-learning, not
necessarily after manual classification. However, it does show that
*especially* low scoring spam should be trained -- definitely, if it
went through below the spam threshold and/or has a low Bayes ratio.
The AWL hit also shows, that you've received spam from that address
before, originating from the same net-block. Given the address, it
clearly also was a scam -- I'd even bet, it was very similar, if not
identical, in text.
Let me take a guess, you did not manually feed the previous one to
sa-learn for training?
--
char *t="\10pse\0r\0dtu\0.@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}
