Re: [Asrg] 2a. Analysis - Spam filled with words

["Jon Kyme" <jrk@merseymail.com>]

> 
> > ...content analysis type things collapse
> > measures made on a number of axes into one metric. 
> 
> Perzactly.  And they have to, ultimately, because the delivery 
> decision itself is binary.  It's basically the same problem that 
> search engines confront when they collapse a high-dimensional 
> document space into a 1-D space like a relevance-ranked list, where 
> the documents below some lower-limit threshold are not displayed.

No. I think you're confusing the dimensionality of the *metric* with the
"ultimately ... binary" (in this application) disposition *decision*. I can
make a spam/ham decision based on the location of a message in some
n-dimensional field ... n can be greater than 1. I may be able to make a
more accurate decision when n is greater than 1. Of course this may be
prohibitively expensive to compute. But any system which makes more than
one metric (if they're anything like orthogonal) available to the filtering
mechanism can allow just this kind of use.





--

_______________________________________________
Asrg mailing list
Asrg@ietf.org
https://www1.ietf.org/mailman/listinfo/asrg