Re: property inference rule

Danny Ayers:
> >> I was seeking for ways  to express this rather basic fact using bare
> >RDF(S)/DAML syntax.
> >
> >There's no way of expressing this in RDF, in RDFS or in DAML; it requires
> >the ability to use variables.  Variables are not part of any of the three
> >representations.
>
> I'm not entirely convinced - for one thing, expressing the relationship
> isn't the same as resolving it, so perhaps something like equivalentTo
might
> do.

I think this might be an excellent idea!
Can you come up with a syntax example? :-)

> For another, if it is viewed as a transformation rather than a
> procedure, no variables are required (XSLT, anyone?).

You are right! This was exactly the reason I started to think about it:
Working on one of my stylesheets I realized that I can easily do it,
but could not come with RDF source code.

> If I'm shot down on
> these points, then there's always TimBL's cwm and log:implies.

Sounds interesting, it would be very interesting to see an N3 expression for
this:

property "C" = property "A" of some subject which is an object of property
"B"

Is this a close guess:
{ :x A :y. :y B :z} log:implies { :x C :z }          ?

How would this look in RDF/XML representation?

Thanks,

--Nikita.

Received on Monday, 12 November 2001 09:15:58 UTC