Re: Action-166 Draft sketch of how to serialize rdf annotation spaces - separate files

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Subject: Re: Action-166 Draft sketch of how to serialize rdf annotation spaces - separate files
Date: Wed, 16 Jul 2008 12:56:29 -0400

> On Jul 16, 2008, at 12:40 PM, Peter F. Patel-Schneider wrote:
> 
> > From: Alan Ruttenberg <alanruttenberg@gmail.com>
> > Subject: Action-166 Draft sketch of how to serialize rdf annotation
> >          spaces - separate files
> > Date: Wed, 16 Jul 2008 02:20:38 -0400
> >
> >> After some thought I have proposed a scaled down version of Rich
> >> annotations based on ideas from Bijan's proposal and a paper by a few
> >> people we know:
> >>
> >>
> http://www.comlab.ox.ac.uk/people/boris.motik/pubs/dhmgh08-metalevel-information.pdf
> >>
> >> Details at:
> >>
> >> http://www.w3.org/2007/OWL/wiki/Annotation_system_2
> >>
> >> -Alan
> >
> > I am puzzled as to why the separation of the serialization into two
> > files.
> >
> > What problems is this split serialization supposed to solve?
> 
> We didn't have any current proposal for serialization of rich
> annotations, so this is intended to move from 0 to 1.

Is the proposal really "rich annotations"?  All I see is regular
annotations plus annotations on entity annotation annotations.

I thought that I had proposed a way to serialize annotations on
annotations, which was to proceed just as in serializing regular
annotations.  This may have only been in a teleconference.

Here I repeat the proposal:

Annotations on annotations are serialized in RDF by "OWL reifying" the
annotation triple and annotating the "reification" node in the usual
way.

> > What benefits does this split serialization provide?
> 
> The intent was to allow, with relatively simple tooling changes, for
> there to be two "spaces" over which to reason, one for the domain, and
> one for the annotations. Each file can be separately fed to a standard
> reasoner, checked for consistency, queried, etc. In that sense it tries
> to meet some of the use cases for annotation spaces, but with what I
> hoped would be considered a conservative proposal.

Certainly O and m(O) can be separately reasoned with, but does the split
actually achieve this?  In the proposal the first file still has quite a
bit of stuff related to the annotations, so it is not really just O!

> > I have a number of other questions, but I don't think that there is
> > > any reason to discuss these other questions until the two-file
> > > question
> > > is discussed.  I don't see any discussion of this in the minutes of 09
> > > July 2008.
> 
> http://www.w3.org/2007/OWL/meeting/2008-07-09#line0434

Sorry, I meant that there wasn't any discussion of why the two-file
solution was reasonable.

> -Alan

peter

Received on Wednesday, 16 July 2008 17:29:26 UTC