Re: evaluating DAML+OIL vs. WebOnt requirements

> > 1) Extending your recipe for strings, can't one express
> > unique URI naming of objects using
> > 
> >   :uniqueName
> >     a daml:UnambiguousProperty;  # each uniqueName denotes 1 object
> >     a daml:UniqueProperty.       # each object has only 1 uniqueName
> 
> yes; how is that different from what I wrote?

Your example used strings (DatatypeProperties).  I applied
it to ObjectProperties (a current requirement for using
UnambiguousProperty), which I think are more likely to be
used for unique names.

> > 2) For ontology management language features, I'd add that
> > DAML+OIL supports the use of other properties (such as
> > Dublin Core) with ontologies, but doesn't give them meaning.
> 
> What do you mean by that? It gives them just as much meaning
> as any other ground fact, no?

The ability to add any property you want is a capability
that we can easily take for granted, but certainly wouldn't
want to lose.

"Glorified comments" is a good description for
daml:versionInfo; we may want to add some more such
properties (somewhat akin to rdfs:isDefinedBy and
rdfs:seeAlso) and/or encourage the use of Dublin Core so
that everyone won't unnecessarily reinvent their own.

> I thought we did split it into two requirements:
>   Annotation/tagging of (whole) ontologies, which is an A requirement, and
>   tagging/grouping, i.e. giving properties to parts of ontologies, which got a B.

A couple thoughts here:

1) Although adding properties to a single object or
statement would be covered by "part"; I think it's a
sufficiently important case to deserve separate
consideration.

2) Adding properties to instances is straightforward; adding
properties to statements (which is what I mean by tagging)
is less clear (particularly if one wants to avoid all of the
other baggage and bloat associated with reification).

Thanks!

	Mike

Received on Wednesday, 30 January 2002 16:26:54 UTC