Confused

[Guy Rixon]

IIRC, the IVOA process for defining bits of data model (and the process is
semi-formaly defined, but I don't have the URL to hand) goes like:

1. Define the new data-model fragement in UML. Churn, iterate until happy.

2. Represent the model as XMI.

3. Send the XMI to some mailing list at iova.net.

4. Discuss, churn, iterate until consensus or until everybody gives up.

5. Assuming consensus on the object model, produce an XML schema to govern
applications of it.

So, yes, it's a very, very structural thing.  In fact, the output is 100%
structural and semantic stuff can only be included as comments.  As you note,
you can't interrogate this kind of model to learn its semantics.

Thing is, we can do this _now_, whereas the higher-level semantic stuff seems
to depends on tools and techniques that we, the service implementors don't
know (saving the presence of the people on this list who _do_ know this stuff
coz they invented it) and which seem a bit vaporous anyway. (OK, 100% and
condensing, but still not quite there yet.)


On Mon, 11 Nov 2002, Tony Linde wrote:

> I think you're right about it being an implementation issue, Guy. I
> think the confusion arises because, having spent twenty years or more
> with 'data model' being an ERM (or similar) diagram produced before
> implementing a database design, I don't see how it can be used as
> 'metadata' or an 'ontology'.
>
> However, I guess if the model is produced using UML and stored as XMI
> then it could be interrogated.
>
> Is this what the 'data model' people are doing? Specifying a data model
> for VOs that can also be interrogated as metadata? In which case what
> standards are being used for the data model format?
>
> I guess this is a useful way of storing structural metadata with some
> relationship information in the links between 'tables' but I wonder how
> much and what type of information cannot be stored in this way. And is
> it important to what we want to do?
>
> I think the biggest potential problem is that it IS structural. An
> ontology or loosely stored metadata does not mandate how the data itself
> is stored. An important point when many projects are working towards an
> interoperable VO but not using or even based on the same implementation.
> A data model may enforce structures that some people do not want to use.
>
> Hmm, what do others think?
>
> Cheers,
> Tony.
>
> > -----Original Message-----
> > From: Guy Rixon [mailto:gtr at ast.cam.ac.uk]
> > Sent: 10 November 2002 16:28
> > To: Tony Linde
> > Cc: semantics at us-vo.org
> > Subject: Re: Confused
> >
> >
> > Tony,
> >
> > perhaps it's an implementation dispute rather than a definition thing?
> >
> > The IVOA data models are now specified to be expressible
> > using UML.  That is, they only contain relationships "is a"
> > and "has a" (and perhaps "owns" if we distinguish composition
> > from aggregation).  The ontological content is restricted, so
> > much of this modelling can be done with existing tools that
> > support OOA.
> >
> > Conversely, if we use lots of relationships, then we have to
> > use specialised tools from recent ontology research; tools
> > that look vaporous right now.
> >
> > I think your definitions are right.  I also think that many
> > VO folk ignore the ontological theory behind their current
> > modelling because it's not necesary to know the thory to use
> > the technique.
> >
> > On Fri, 8 Nov 2002, Tony Linde wrote:
> >
> > > I've seen comments saying that we should use metadata instead of
> > > ontologies, data models instead of ontologies, data models
> > instead of
> > > metadata etc.
> > >
> > > So what is the difference? And what do people mean by these terms?
> > >
> > > For myself:
> > >
> > > Data is information stored electronically.
> > > Metadata is data which describes some other data.
> > > Dataset is a collection of data organised by some principle. Data
> > > model is an ERM of a dataset which describes its underlying
> > > construction.
> > >
> > > Ontology is all of the terms and semantic relationships
> > used in some
> > > domain discourse (which may be discovered and stored
> > digitally but not
> > > necessarily).
> > >
> > > How about you?
> > >
> > > Cheers,
> > > Tony.
> > >
> > > __
> > > Tony Linde                       Phone:  +44 (0)116 223 1292
> > > AstroGrid Project Manager        Fax:    +44 (0)116 252 3311
> > > Dept of Physics & Astronomy      Mobile: +44 (0)7753 603356
> > > University of Leicester          Email:  ael at star.le.ac.uk
> > > Leicester, UK   LE1 7RH          Web:    http://www.astrogrid.org
> > >
> > >
> > >
> >
> > Guy Rixon 				        gtr at ast.cam.ac.uk
> > Institute of Astronomy   	                Tel: +44-1223-337542
> > Madingley Road, Cambridge, UK, CB3 0HA		Fax:
> > +44-1223-337523
> >
>
>

Guy Rixon 				        gtr at ast.cam.ac.uk
Institute of Astronomy   	                Tel: +44-1223-337542
Madingley Road, Cambridge, UK, CB3 0HA		Fax: +44-1223-337523