RE: Comment: ISSUE-70

Thank you, Sean, that is much clearer.  
What are the arguments against making skos:broader irreflexive?

--jody

-----Original Message-----
From: Sean Bechhofer [mailto:sean.bechhofer@manchester.ac.uk] 
Sent: Wednesday, February 13, 2008 4:20 AM
To: Deridder, Jody L
Cc: public-swd-wg@w3.org
Subject: Re: Comment: ISSUE-70


On 12 Feb 2008, at 18:01, Deridder, Jody L wrote:

>
> Thank you, Sean.
>    I'm still, however, confused.  It seems to me that these examples 
> can be reasonably considered as sets and subsets.
>
> The set of fairy tales contains the set of Cinderella stories as well 
> as the set of Rumpelstiltskin stories.  The set of Cinderella stories 
> is a subset of the set of fairy tales.

Maybe, but think of this example:

mountain ranges
   Andes
   Himalayas

Now here, the Andes is an instance of a mountain range rather than being
a subset. Maybe you consider that the Andes contains (errr, quickly
consults the web...:-) Pomerape, Monte San Valentin, Alpamayo etc, but
these are not then mountain ranges, they're mountains. So it's not quite
subset/superset inclusion that's going on here.

> Likewise, regiments are a subset of battalions, which are in turn a 
> subset of military divisions and so on.

But if we look at the later examples:

nervous system
   central nervous system
     brain

The brain is not a kind of central nervous system, e.g. the set of
brains is not contained in the set of central nervous systems. Its a
*part* of the central nervous system. Here we have a broader partitive
relation.

Similarly:

armies
   divisions (military)
     battalions
       regiments

Would you say that a regiment was a kind of army? I think probably not
-- armies are composed of divisions, which are composed of battalions
and so on. Again, we have a partitive relation.

What I'm trying to say here is that we shouldn't simply base our
interpretation of broader on a subclass relationship as there are other
ways in which it's used. SKOS is intended to be used to represent
general Knowledge Organisation Systems, so we have a much "looser"
interpretation of broader. If you really have a situation where you know
you can interpret all your concepts as classes or sets of individuals,
and you have a super/subclass relationship, then a more appropriate
representation might be OWL [1], which then brings advantages such as
inference and consistency checking (but also some costs).

Now, one could argue that the examples above lend weight to the argument
that broader should be irreflexive. However, I think that's a
*different* (but also important) argument.

Cheers,

	Sean

[1] http://www.w3.org/2004/OWL/
--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@manchester.ac.uk
http://www.cs.manchester.ac.uk/people/bechhofer

Received on Wednesday, 13 February 2008 12:58:44 UTC