Re: RDF-ISSUE-25 (Deprecate Reification): Should we deprecate (RDF 2004) reification? [Cleanup tasks]

On Sat, 2011-04-09 at 12:25 +0200, Ivan Herman wrote:
> William,
> 
> On Apr 9, 2011, at 12:06 , William Waites wrote:
> 
> > * [2011-04-09 11:51:11 +0200] Ivan Herman <ivan@w3.org> écrit:
> > 
> > ] I must admit I do not understand what this has to do with 
> > ] reification. Yes, the model you describe is fairly common
> > ] and, to take an even more widely used example, it is the
> > ] same pattern as the one used for a foaf:Person.
> > ] 
> > ] I seem to miss something here.
> > 
> > The difference being that a foaf:Person is a thing in some sense, and
> > an org:Membership really wants to be a relationship. And relationships
> > are meant to be expressed as predicates except that we can't do that
> > here because it wants four arguments instead of two. What would be the
> > natural way to write down a foaf:Person in prolog? What would be the
> > natural way to write that this person is a certain type of member of
> > something?
> 
> It is difficult for me to really appreciate what you say because I do not know the details of that particular ontology. I am That specific issue put aside, I understand that expressing n-ary predicates are a bit complicated; see also:
> 
> http://www.w3.org/TR/swbp-n-aryRelations/
> 
> I still fail to understand how this relates to the specific issue of reification. Both the example you gave, as well as the foaf structures, but also the examples in that Note, seem to work without reification, ie., this does not seem to be a use case for it...

Just guessing, but I think the connections are:

(1) What William is talking about is (I think) technically also called
"reification".   That is, one way to handle n-ary relationships in RDF
is the reify them.  So, the short answer to William is "No, ISSUE-25 is
only about the the reification of RDF triples.")

(2) The mechanisms for making RDF reification of triples efficient are
the same as one might use for efficiently handling reified
relationships: that is, you present or consume an RDF view of some
internal data structure.   When you do this, on the consumption side,
you have some complications in handling streams.

While I'm writing about ISSUE-25, I should note that RIF-in-RDF [1] is
another RDF reification mechanisms, since RIF is a superset of RDF.  

For completeness, I'll note that OWL 2 came near this issue, and
provides RDF negation through something that looks like reification,
with Negative Property Assertions [2].

    -- Sandro

[1] http://www.w3.org/2005/rules/wiki/RIF_In_RDF
[2] http://www.w3.org/TR/owl2-primer/#Object_Properties


> Ivan
> 
> 
> > 
> > Or to put it another way it might have been possible for the authors
> > of org to do the same thing by making a memberOf predicate and then
> > using sub-predicates to refine the idea. There might be many such
> > sub-predicates as they would be parametrised by time periods. This is
> > kind of like curried predicates if you will. Even though it might be
> > more natural to think of things this way, predicate explosion, the
> > fact that some stores do not like enormous amounts of ad-hoc
> > predicates, and the general lack of subproperty reasoning would make
> > this way actually worse than the reification they use.
> > 
> > Does that make more sense?
> > 
> > -w
> > -- 
> > William Waites                <mailto:ww@styx.org>
> > http://river.styx.org/ww/        <sip:ww@styx.org>
> > F4B3 39BF E775 CF42 0BAB  3DF0 BE40 A6DF B06F FD45
> > 
> 
> 
> ----
> Ivan Herman, W3C Semantic Web Activity Lead
> Home: http://www.w3.org/People/Ivan/
> mobile: +31-641044153
> PGP Key: http://www.ivan-herman.net/pgpkey.html
> FOAF: http://www.ivan-herman.net/foaf.rdf
> 
> 
> 
> 
>

Received on Saturday, 9 April 2011 12:17:17 UTC