W3C home > Mailing lists > Public > public-owl-wg@w3.org > February 2009

Re: ACTION-264: Discuss imports with Tim Redmond.

From: Timothy Redmond <tredmond@stanford.edu>
Date: Mon, 16 Feb 2009 12:03:14 -0800
Cc: public-owl-wg@w3.org
Message-Id: <8F1B5FB4-BB9E-4766-8813-AA0571FB9871@stanford.edu>
To: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>, Bijan Parsia <bparsia@cs.manchester.ac.uk>, Alan Ruttenberg <alanruttenberg@gmail.com>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>

I wanted to add one more comment relevant  to use cases.  In my  
experience, the most common reason why ontology developers do not  
import by name is that the ontology being imported can't be found by  
using its name.  Often the ontology developers feel that they have no   
choice  because they do  not  have any control over the ontology that  
they are importing.

The next most common case is that the import just happened to be  
written that way. The PATO case I mentioned  in my last note occurred  
in an ontology being  developed by NCI.  The imports link was simply  
lifted from the obofoundry web page even though importing by name  
would have worked perfectly fine.  One version of the birnlex also had  
this feature - one ontology in  the imports closure happened to be  
imported in two different ways.

The ontology version concept in OWL 2 will also lead to some ambiguity.

Finally, I am seeing more ontologies being developed for agent based  
or tomcat based applications.  Often in these cases the import will  
have the form

	imports http://localhost:8080/....


On Feb 16, 2009, at 5:43 AM, Ian Horrocks wrote:

> Just to be clear, I assume we are talking about recommendation in  
> the usual sense and not in the W3C sense. Even in this case, I think  
> that recommending this particular solution may be too strong.  
> Suggesting it should be OK though.
> Ian
> On 14 Feb 2009, at 00:56, Timothy Redmond wrote:
>> I am not very knowledgeable about XML catalogs but they do look  
>> like exactly the right thing.  In fact it looks like the suggestion  
>> goes beyond my original query.
>> Without the recommendation though, different tools will probably  
>> end up using different solutions.  While not fatal this is awkward  
>> for sharing between different systems.  Even Protege 3 and Protege  
>> 4 have some of this awkwardness already.
>> So if XML catalogs make sense I favor the recommendation.
>> -Timothy
>> On Feb 12, 2009, at 1:42 PM, Bijan Parsia wrote:
>>> Sorry not to delve into to the emails (too much on my stack at the  
>>> moment), but I'm unclear why Protege adopting something like XML  
>>> Catalogs doesn't solve everything without changes to the current  
>>> spec. Indeed, forget "like", just use XML Catalogs.
>>> P4 could even export a catalog to a zipped directory which  
>>> maintains the mappings.
>>> I'd be happy with us recommending this, even.
>>> Cheers,
>>> Bijan.
Received on Monday, 16 February 2009 20:03:51 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:42:09 UTC