Re: Re: Higher-arity to RDF binary

On Thu, May 20, 2010 at 14:59, Chris Dollin <chris.dollin@epimorphics.com>wrote:

> On Thursday 20 May 2010 01:20:54 pm Jitao Yang wrote:
> > for example:
> > "... the formula DescPr(*d, p, o*) in *L*, asserting that object *o* is a
> > value of property *p* in description *d*. ..."
> >
> > and we have:
> > DescPr(*d, p, o*)
> > DescPr(*d1, p, o1*)
> >
> > the above formula could be represented by RDF like:
> >
> > :d
> >        a                           :DescPr_description ;
> >        :has_property       :p .
> >
> > :d1
> >        a                            :DescPr_description ;
> >        :has_property        :p .
> >
> > :p
> >        a                          :DescPr_property ;
> >        :has_value           :o ;
> >         :has_value           :o1 .
>
> That would be wrong, because if d has_property p which has_value
> o, it /also/ has_property p and value o1. Similarly for d1. Hence the
> RDF representation does not encode the two formulae given; it
> encodes /4/ formulae, dpo, dpo1, d1po, d1po1.
>
> > The class DescPr_description and DescPr_property are defined in RDFS:
>
> Only if you use "define" in a very weak sense. Admittedly that's the
> best you can do in RDFS (and OWL), but what you can't do is provide
> any non-comment specification of how these properties are something
> to do with the formulae.
>
> > of course, the triples can go back to the logical representation, in the
> > above example.
>
> No; that representation doesn't allow you to recover all /and only/ the
> original formulae.
>
> > and there are errors in the above RDFS, since based on the above RDFS: d
> =
> > d1,
>
> No. RDFS doesn't impose identity constraints like that.
>
> > but in fact no.
> >
> > Use your idea:
> >
> > case3:
> >
> >    [a Formula; relation p; on d; value o].
> >    [a Formula; relation p; on d1; value o1].
> >
> > case 1:
> >
> >    p assertion [on d; value o], [on d1; value o1]
> >
> > but, how to translate your representation to Turtle?
>
> They're pretty much in Turtle; you need to choose prefixes
> for the terms I've written as "relation", "on", "value", and
> "Formula", but that's about it.
>
> > because your representations are not triples.
>
> Consider
>
>  p assertion [on d; value o], [on d1; value o1]
>
> Let's pick namespaces and prefixes to illustrate:
>
> @prefix eh: <http://rdf.epimorphics.com/chris/examples#>.
> @prefix some: <http://example.com/rdf#>.
>
> some:p eh:assertion
>    [eh:on some:d; eh:value some:o]
>    , [eh:on some:d1; eh:value some:o1]
>    .
>
> if I understand correctly? The above representation could be translated :

some:p               eh:assertion             _blanknode1 ;
_blanknode1      eh:on                       some:d  ;
_blanknode1      eh:value                   some:o  ;
some:p               eh:assertion             _blanknode2 ;
_blanknode2      eh:on                        some:d1 ;
_blanknode2      eh:value                    some:o1 .

but it seems lost the connections between b, p and o?


> That's all triples, using Turtle syntax for multiple objects
> (ie: S P O1, O2, O3 abbreviates S P O1. S P O2. S P O3)
> and blank nodes ([...]) with multiple property-values
> (S P1 O1; P2 O2 abbreviates S P1 O1; S P2 O2).
>
> Not counting prefixes or extensions for quads, if you can
> write it in Turtle, it's triples.
>
> --
> RDF is not /the/ answer. RDF is /an/ answer.                         -
> Arcadian
>
> Epimorphics Ltd
> Registered address: C/O Robson Taylor, Froomsgate House, Bristol
> Registered number:  7016688
>
> -----------------------------------------
> --
> RDF is not /the/ answer. RDF is /an/ answer.                         -
> Arcadian
>
> Epimorphics Ltd
> Registered address: C/O Robson Taylor, Froomsgate House, Bristol
> Registered number:  7016688
>
>