Re: discussion of ISSUE-16 - was draft of minutes from telecon of 2 July now available

On 3 Jul 2008, at 23:55, Peter F. Patel-Schneider wrote:

> Hmm.
> I expect that the minutes are a factual representation of the
> discussion, but some of the discussion does not appear to me to be
> correct.
> Section 2.1.1: ISSUE-16 Entity Annotations
> The issue raiser was Jeremy Carroll.

Good point, but I guess that it didn't affect the subsequent discussion.

> The issue itself is about the ability to annotate an entity annotation
> (because it is an axiom) but not other annotations (because they  
> are not
> axioms).

I think that Boris did correctly summarise the issue, but that it got  
a bit scrambled in scribing -- to be fair to Evan, it wasn't easy to  
I have amended the minutes to be more consistent with my recollection  
of what was actually said:

Boris Motik: problem - you can annotate entity annotations, because  
they are axioms, but not other kinds of annotation, because they are  
not axioms ←
... peter proposed various solutions, including one where annotations  
could contain a set of other annotations

> I did put forward a group of proposals to close the issue in
> a
> subset of which allow annotations on annotations, which requires  
> only a
> very small change to the functional syntax.

The agenda pointed to this email.

> The situation in the RDF
> syntax would be a bit more complex, but I believe that simply using  
> the
> same treatment as for single-triple axioms would work (the annotated
> annotation would be reified and the annotations on "hung" off the
> reification node).
> There was support for one of my proposals (2a), which is the one  
> that I
> also prefer.

Your (2a) solution was one of the ones under consideration. The  
alternative proposed by Boris -- making all kinds of annotation be  
axioms -- was what he was referring to when he said "having an axiom  
that contains another axiom is hard in RDF". We were also concerned  
that reifications of reifications would be quite nasty -- there is  
already some concern about the overhead for parsers of having to  
handle single reifications, and a debate as to whether the "base  
triple" should also be included in the RDF serialisation. These  
problems/debates might be made worse by the possibility of having to  
deal with arbitrarily nested reification.


> peter
> PS:  My proposal was very different from the one put forward by Boris.

Received on Monday, 7 July 2008 12:25:31 UTC