Re: Another approach to Nary relations

Dear Natasha,

Sorry for being late in sending you my reply message.

Natasha Noy:
> Up until the reification and probability, it seems you are basically
> using pattern 2 from the WD "as is", aren't you? Except that you also
> allow for there being only one or two components in the relation.
> Pattern 2, however, while presented as to solve a problem for relations
> with arity > , can indeed be applied to relations with fewer arguments.
> Perhaps, it's a point worth making under "Considerations".

Yes, that's my point.
I dont't think it's is natural to express binary relations in the
conventional way
while expressing  unary or 3(or more)-ary relashionships in the "predicate
centric" way.

If we adopt the predicate centric method in all cases, we shall have
plenty of "legacy"  in the conventional way.
However, we can provide translation tools to deal with them.

> Reification and probability, however, adds an extra level there that is
> probably more appropriate if indeed you want to treat uncertainty
> property, but is not really a subject of a note on n-ary relations.

Right.

However, with the predicate centric approach, we can introduce an syntax
sugar for reification
which will decrease the complexty accomapnied with reification.
In a word, we can REIFY the predicate node.
I'm writng a document on that.

So, it would not be a bad idea to refer to the reification issue in
the working draft, I think.

>  It
> seems, however, that it may be a good idea to write a separate note
> describing design patterns for representing uncertainty. What do others
> think? I believe, adding this note to the set of OEP documents would be
> a very good idea, and Yoshio has essentially already started on that.

And it would be my pleasure if I can discuss this stuff with people here
/ and out in the world, and I can contribute in forming the best practice
in in expressing uncertainty in  RDF.

Best,
Yoshio Fukushige
fuku@w3.org

Received on Monday, 30 August 2004 03:33:54 UTC