W3C home > Mailing lists > Public > public-owl-wg@w3.org > June 2008

Re: Question about problems with top/bottom property

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Mon, 2 Jun 2008 13:46:53 +0100
Message-Id: <2395DDA5-2259-4843-BACE-4D39103C3792@cs.man.ac.uk>
Cc: Alan Ruttenberg <alanruttenberg@gmail.com>, Boris Motik <boris.motik@comlab.ox.ac.uk>, 'Michael Schneider' <schneid@fzi.de>, public-owl-wg@w3.org
To: Ivan Herman <ivan@w3.org>

On 2 Jun 2008, at 13:16, Ivan Herman wrote:

> Alan Ruttenberg wrote:
>> Is there any reason not to include bottom role? There is a  
>> debugging benefit to computing equivalentProperty to bottom role.
>
> I must admit I do not understand what you mean here.
>
> In general, I would like to understand the clear benefit the top  
> and bottom role would bring to OWL users. At the moment, it is  
> unclear to me.
[snip]

We have had extensive discussion on this, so perhaps as summary is due.

 From a UI perspective, Top and Bottom properties add symmetry (i.e.,  
analogues to Thing and Nothing) and thus a more uniform UI. For  
example, right now, it is rare (unknown?) for reasoners to report  
unsatisfiable properties (which do occur). A natural way to report  
this is to show them as equivalent to or subsumed by a Bottom  
property (in analogy with how unsatisfiable classes are handled).

Similarly, I find users adding an artifical top property (or asking  
for one) just to help organize their properties. (I find this a bit  
odd, personally, but that's what it is.)

 From an expressiveness point of view see:
	http://www.w3.org/2007/OWL/wiki/Universal_Property

In general, TopProp (my new favored name :)) allows one to express co- 
existence constraints *without* specing a particular relation between  
the two entities. For example, you might wish to express that if  
there is a disease occurrence then there is a cause (germ, poison,  
trauma, genetic defect) without necessarily having a "local top"  
causal property (i.e., a generic caused by):

	DiseaseOccurrence sub (someValuesFrom owl:universal owl:Thing)

(or some more specific class of causal agents).

	DiseaseAfterTraumaOccurrence sub (someValuesFrom owl:universal Trauma)

(Where the way the trauma causes the disease might be unspecified or  
one of a number of disjoint mechanism).

In this case, you can capture the structure by other means (including  
simulating the TopProp). But it does seem more direct and flexible.

Cheers,
Bijan.
Received on Monday, 2 June 2008 12:44:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 2 June 2008 12:44:48 GMT