Re: closing semantic issues (reification)

Pat Hayes wrote:
> 
> >
> >  > True, and I wish it did address reification. For reasons that are
> >>  still opaque to me, the treatment of reification that I originally
> >>  offered was rejected by the WG, but until someone can tell me *what*
> >>  was wrong with it, I am somewhat at a loss as to how to proceed.
> >
> >I have told you several times now (including ftf on the
> >walk to the bar in California), but I guess I'll say it once more:
> 
> You told me that I had it wrong, but you didn't tell me what was right :-)
> 
> >In your model theory, the subject of the sentence
> >"Mary hit the ball" is a word starting with the letter "M".
> >In RDF-as-deployed, the subject of that sentence is a girl.
> >
> >Slightly more precisely, given
> >       <Mary> <hit> <aBall>.
> >its reified form includes
> >       _:statement rdf:subject <Mary>.
> >
> >note that the reified form doesn't quote <Mary>.
> 
> Well, OK, but that is completely at odds with everything the M&S says
> about reification.

Yes, I agree, it's horribly botched. I'm not arguing that
it *should* be this way; only observing that it has been widely
interpreted this way.

Maybe we can convince
the implementors that have implemented it this way that it's
a bug. But it's a widely deployed bug. Perhaps not a lot
of applications depend on this behaviour, and it's
feasible to "fix" the bug; i.e. redeploy the implementations.


> A reified statement is supposed to be a
> reification *of a statement*, ie of a triple, right?

Well, yes, in the applications I want to build, that's
the sort of reification that would be useful.

> In your account
> here, the reification of the triple
> 
> <mary> <hit> <ball> .
> 
> has exactly the same content and meaning as the original triple. If
> this is the case, what is the point of using reification at all?

I have no idea why anybody would want this form of reification.

> All
> one has done is to provide another way to write the same thing.
> 
> [Later. Oh, I see what the answer might be: same proposition, but not
> same assertion. Reification here isn't thought of as a metalevel
> assertion, but rather as a way of sidestepping assertion. The point
> being that the unreified triple is asserted in any graph in which it
> occurs, but the reification is not asserted. Is that the idea?]

Not for me.


> Tell you what, rather than get into an immediate debate about this, I
> will try to write a short summary of various views on what
> reification  means, which that will give us something to refer to. (I
> now have four of them. ) I was planning to do this last week but got
> zapped.
> 
> Pat

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Wednesday, 7 November 2001 10:20:21 UTC