Where to start (was: Ontology for Dummies)

[Sean Bechhofer]

If this discussion is proving uninteresting or irrelevant to the mailing
list, please let us know and we can take it elsewhere....

On Fri, 4 Oct 2002, Bernard Vatant wrote:

> * Sean Bechhofer
> > Topic Maps and OWL (hereafter referred to as TM and OWL to save
> > typing) *are* different. This is not to say that one is "better" than
> > the other though -- they are *different*, and are suitable for
> > different purposes. TMs provide a kind of model of "back of the book
> > indexes", a way of indexing particular occurrences of topics or
> > subjects. In my opinion, they are not a framework for representing
> > ontologies (** see below).
>
> This is controversial, even TM folks are not quite clear on that.
> TM are a framework for representing taxonomies and semantic networks.
> What is missing (so far) is the constraint and inference layer.

Ok. I suppose the issue here is that I think that inference is crucial if
we're going to get added value from machine processing (and really have a
*Semantic* web).

> > The crucial point with OWL is that we have a well-defined notion of
> > what it means when we say:
> >
> > telephone -> all colour black
> >
> > i.e. all telephones can *only* have the colour black.
>
> Anyway. I don't see why TM could not express that by using specific topics and
> associations.
>
> Define a topic representing the class "telephone"
> Define a topic representing the concept "colour'
> Define a topic representing the colour "black"
> Define a topic representing the concept "class"
> Define a topic representing the association type "class_colour"
>
> Every one of the above being defined as a Published Subject if possible.
>
> Then define an instance of the association-type "class_colour"
> where "telephone" plays the role of "class" and "black" plays the role of "colour"
>
> class_colour (class: "telephone" ; colour: "black")
>
> And, at will, constrain the domain of validity of this assertion by a convenient scope,
> like "1930, USSR" ;-)
>
> Is there less semantics in the TM expression than in the OWL expression?
> Let alone the capacity of the latter to draw inferences ...

I'm afraid I don't quite follow this, and don't see where the semantics
are here. How do I know (without you having to tell me extra stuff about
the interpretation of your particular association types) that the
assertion above should be interpreted as the given constraint? I.e. that
class_colour restricts the class to having the particular colour. Is this
how I *always* interpret an association type? And if not, how do I ensure
that this information is conveyed?

As a more complicated example (which is again, I'm afraid,
non-astronomical), consider the class of people who have a son and all of
who's children are Lawyers. In OWL, I'd say something like:

PersonWithSonAndAllLawyerChildren =
 Person AND
 all child Lawyer
 some child Male

How would I represent this information using TMs? And how do I then ensure
that I can tell that the class:

PersonWithSonAndDaughterBothLawyers =
  Person AND
  exact 2 child
  some child (Male AND Lawyer)
  some child (Female AND Lawyer)

is a subclass of the first one, e.g. any instances of
PersonWithSonAndDaughterBothLawyers are necessarily instances of
PersonsWithSonAndLawyerChildren (which then allows me to do things like
induce a taxonomy from my definitions).

Cheers,

	Sean

-- 
Sean Bechhofer
seanb at cs.man.ac.uk
http://www.cs.man.ac.uk/~seanb