W3C home > Mailing lists > Public > semantic-web@w3.org > September 2006

Re: Inheriting property restrictions

From: ben syverson <w3@likn.org>
Date: Thu, 7 Sep 2006 12:26:40 -0500
Message-Id: <4F7AEC54-5918-4930-AABF-FE9B49143586@likn.org>
To: semantic-web@w3.org

It has nothing to do with the model; it's just that in my case, a  
union of property restrictions would result in a simpler user  
interface. An intersection means I have to alert the user and make  
sure they really know what they're doing when they have overlapping  
property restrictions.

Part of the issue is that OWL ontologies are typically designed by a  
single person or a cohesive group, whereas my application will allow  
for distributed/wiki-like ontology editing. I'm anticipating  
scenarios where someone adds a restriction but hasn't thought it  
through -- which would be easy to correct with a union.

Take for example, a class named "MacintoshComputer." User A restricts  
MacintoshComputer's "processorType" to allValuesFrom "PowerPC." User  
B comes along and wishes to expand the restriction to include "i386."  
I can imagine that from a user's perspective, they might think "oh,  
I'll just add another restriction." What User B will end up having to  
do is challenge User A's assertion, and then make a new assertion  
that MacintoshComputer's processorType is allValuesFrom  
MacintoshProcessor, and then define MacintoshProcessor as the union  
of PowerPC and i386. (Ideally you would use an enumerated class with  
only those PowerPC and Intel processors actually used in Macs, but I  
don't support enumerated classes in likn...)

It's no big deal -- just something I have to make people aware of.

- ben syverson
likn.org



On Sep 7, 2006, at 11:00 AM, Jon Hanna wrote:

> ben syverson wrote:
>> Thanks Paul!
>> I guess it's as I feared -- multiple property restrictions result  
>> in  the intersection of the restrictions, rather than the union.
>
> Why fear this? Can you think of a real-world counter-example  
> (whether concrete or abstract)? Seems like things are being  
> modelled perfectly sensibly.
Received on Thursday, 7 September 2006 17:27:05 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:53 UTC