On Tue, Apr 10, 2012 at 08:52:08AM +0200, Antoine Isaac wrote:
> >Tom: The difference is that an Ontology describes something in the open world context; it
> > creates a cartoon vision of things in the world -- a conceptual universe.
> > An Application Profile describes the Description Set, in a closed world
> > sense, defining constraints that can be validated.
...
> Let's not make too clear-cut statements here. A lot of papers in the
> "ontology" literature deal with constraints that are beyond the shortcomings
> of OWL with respect to expressing constraints. I could mention formal-heavy
> stuff like the foundational ontology DOLCE. But even an ontology like SKOS
> has a couple of constraints. It's just that they couldn't be expressed in
> OWL. An even an OWL ontology may be able to express constraints that match
> with what you expect to find in an AP.
Can we say that ontologies constrain a conceptual universe ("Things"), and
application profiles constrain a data representation ("Data")? The constraints
in SKOS were about the SKOS "model", not about data created in accordance with
that model.
> Another point mentioned in the minutes: as good practice an ontology
> shouldn't re-define elements from an existing ontology, as APs do sometimes
> for the vocabulary they re-use.
Hmm - APs should really only "re-label" URIs, and those re-labelings should not
contradict or extend the semantic scope (meaning) of the URIs. I suppose re-labeling
is a form of re-defining, but it is not one that affects the machine processing of
the URIs.
...
> And I'm not expecting to be challenged on the content here: I just want us to
> avoid making bold statements on what an ontology is or is not. If we do, we
> should always keep in mind some specific context, such as "ontology as
> enabled by the 2011 OWL spec" ;-)
Agreed. However, it is perhaps useful to contrast "open-" and "closed-world"
constraints.
Tom
--
Tom Baker <[log in to unmask]>
|