Re: comments on OWL-lite: meaning of 'required' values

Dan Brickley wrote:

> There are a couple of ways in which 'required' might be understood.
> In XML-based schema languages, "required" is often "required to be
> present in some XML document / description". Others use "required" in the
> sense of "it isn't a member of class Foo unless it has a value for the
> 'bar' property".
> For example an XML schema-like approach might require that each concrete
> description of a member of the class eg:Person includes a value for their
> eg:biologicalParent property.  This is different to saying something
> like "all people have parents", since the latter doesn't care about the
> content of specific XML instances that describe people. Implementors seem
> to want both.
> This is one of the cultural mismatches between some of the XML and RDF
> work, so might be worth adding some clarifying text to the OWL-Lite
> overview.

Good point. Not only XML vs RDF have different interpretations here, but the 
same different interpretations can be found among knowledge representation 
systems. I agree it's worth adding some clarification on the intended 
semantics here.

> One other comment re
> "Individuals - can be given names or can be anonymous"
> This wording suggests that nameless-ness is an intrinsic characteristic of
> the individual, rather than a characteristic of some description of that
> individual. Two XML documents might partially describe the same
> individual; one description might mention a URI name while the other
> mightn't. Saying "individuals can be... anonymous" encourages the
> expectation that there is a class of nameless individuals.

Again good point. I'll include both in the annotated version for the F2F.
(I intend to make those annotations a life document on the Web, but due to a 
fatal design flaw in creation there are only 24 hours in a day...)


Received on Tuesday, 26 March 2002 15:24:14 UTC