Re: abstract syntax and RDFS

From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
Subject: RE: abstract syntax and RDFS
Date: Thu, 23 Jan 2003 10:38:03 +0100

> > > (b) used a new concept annotationPropertyID for there first URI
> > > (and further points in another message to come).
> >
> > I don't see any particular purpose in this.
> >
> 
> On later reflection, I think it is worth, requiring the first URIref to be
> declared as either an owl:ObjectProperty or an owl:DatatypeProperty (for the
> two different sorts of [aA]nnotation).

This is absolutely the wrong way to go.  

> The advantages are:
> - uniformity of the rule that you must declare all urirefs used
> - somewhere to hang annotations on annotations
> e.g.
> <owl:DatatypeProperty rdf:ID="example">
>   <rdfs:comment>
> This is used to annotate classes and properties to show an example use.
> Typically its object is an XML Literal.
>   </rdfs:comment>
> </owl:DatatypeProperty>
> - permits the use of properties such as rdfs:comment or user defined
> properties in both annotations and the ontology itself.
> 
> The disadvantages are:
> - to retain semantic clarity it is necessary to prohibit all further
> constraints on such properties (i.e. no range, domain, restrictions,
> Functional, InverseFunctional, Symmetric, Transitive constraints)

Precisely.  Annotation properties are outside of OWL and should remain so.

As well, some annotation properties have characteristics (such as
dc:author) that mean that they cannot be either owl:ObjectProperty or
owl:DatatypeProperty.

> Jeremy

peter

Received on Thursday, 23 January 2003 09:14:14 UTC