Re: comments on RDF mapping

On Oct 30, 2007, at 6:01 PM, ewallace@cme.nist.gov wrote:

> For example, the last language constraint described above would  
> mean that we
> couldn't use property chains to describe properties that are  
> derived datavalues.
> For me, that was the primary value in property chains.

BTW, Evan, meant to ask you what you consider the correct behavior  
for property chains  on properties that have mixtures of objects and  
data values. Suppose we have:

p < q o r

and:

Class A, instances a1, a2, a3

a1 p a1
a1 r a2
a1 r 5
a2 q a3
a2 q 7

What do you expect a1.p to be inferred to be? How to handle a1 r 5,  
as 5.q doesn't make sense.

-Alan

> I don't understand why
> a simple non-logical annotation vocabulary as suggested recently by  
> Vit Novacek
> in [1] wouldn't achieve the original goals without this high cost.