Re: Reification: naive question

From: "Enrico Franconi" <franconi@cs.man.ac.uk>

> Sorry if this question looks naive, or if you had on this already a
> big discussion in the past (which is likely anyway).
>
> Suppose that a unique id is associated to each triple. This could be
> either implicit (i.e., generated internally like OIDs in O-O data
> models) or explicit (if you want to mention it later).
>
> The additional (explicit) ID serves as a reference in other triples
> willing to state something on it (as a foreign key). I understand that
> this is the spirit of reification.

> For the triples where the ID is implicit, the syntax wouldn't change,
> and the semantics could be the standard MT already devised.
>
> If the current MT wants to ignore reification (as it correctly does),
> then it should just ignore the presence of those additional IDs. This
> makes a lot of sense since the semantics of reification is still
> unclear, and a lot of work should be done. A future extension of the
> MT (based supposedly on HOL) could then take IDs into account.
>
> This is more or less in the spirit of Nejdl's proposal.
>
> - do you have the feeling that this would solve all the problems (on
>   expressiveness and on semantic clarity) of reification?  [please,
>   note the naivete of the question :-)]
>
> - Would this be an impossible addition to the rdf standard syntax?

Actually there is the idAttr allowable on a property element .. see section
D production 6.12 of the revised syntax:
<http://www.w3.org/TR/2001/WD-rdf-syntax-grammar-20011218/#section-Grammar>
But I'm not sure whether it is practical to used that for your purpose.

But isn't it better just to keep piling more and more restraints on a
reified statement untill it can only be satisfied by one instance of the
triple?  See my example at the bottom of the following post:

http://lists.w3.org/Archives/Public/www-rdf-logic/2002Feb/0031.html

Seth Russell

Received on Wednesday, 6 February 2002 11:05:56 UTC