W3C home > Mailing lists > Public > public-webont-comments@w3.org > September 2003

Re: why not collapse sameAs, equivalentClass...and other questions

From: Jim Hendler <hendler@cs.umd.edu>
Date: Wed, 24 Sep 2003 07:48:37 -0400
Message-Id: <p05200f40bb972d63dfdd@[10.0.1.4]>
To: graik@web.de, public-webont-comments@w3.org
Raik-
   A number of the questions you ask were discussed at length by our 
WG and info about what we decided and why can be found on our issues 
list, which is at http://www.w3.org/2001/sw/WebOnt/webont-issues.html
  below I'll give quick answers to your questions with pointers to longer ones
  -JH


At 10:48 AM +0200 9/24/03, Raik Gruenberg wrote:
>Hi all,
>
>since a year or so (on and off) I have been working on a java library for the
>data mining and creation of semantic web documents (you can plug in any other
>ressources too). First of all, I used that for my own data integration
>problem but it is now very general and could be useful for others as well.
>More about that later.
>It's been quite a while since I have visited the OWL pages, and I was
>surprised by the advance and the explicit call for implementator's comments.
>Sorry about the delayed response, but there are two or three things I always
>wanted to know and never dared asking...
>
>1) More a practical question: Where are the real (preferable N3) definitions
>of the current OWL items. I found a lot of helpful documents but no file
>where one can actually check whether e.g. equivalentClass is a sub-property
>of sameAs or not.
>

Appendix B of the OWL Reference (http://www.w3.org/TR/owl-ref/#appB) 
is an RDF Schema for OWL.  It is also available on line at
  http://www.w3.org/2002/07/owl



>2) What are the plans for N3 vs. XML-based Syntax?

I'm not quite sure I understand the question -- OWL is in RDF 
compliant with the new RDF Specs, and N3 works for any RDF thus OWL 
can easily be written in N3.  There's no plans for a specific OWL N3 
document since it is easily produced from the XML-based RDF syntax.

>
>3) My library decides at runtime what "plugin" to use for reading / writing a
>given address. Have you come up with naming conventions ala "a file in N3
>syntax should be called *.n3" etc.?

This is actually more complicated than you might think, but the 
bottom line is that the WG decided not to create a new media type for 
OWL.  This is discussed in the issues list as 
http://www.w3.org/2001/sw/WebOnt/webont-issues.html#I5.13-Internet-Media-Type-for-OWL


>
>4) I am now implementing the automatic follow-up and collapsing of Elements
>related by owl:sameAs. Why is owl:sameAs not also used for Classes and
>Property definitions? Are equivalentClass and equivalentProperty, at least,
>subPropertyOf owl:sameAs? As I understand, each owl:Class definition will
>have a property "owl:type owl:Class" and each Property definition will have
>one "owl:type owl:Property". It is hence quite clear what to expect on the
>other side (although it helps to have a more restricted owl:range).
>
>In my implementation the java classes for Class and Property definitions are
>both derrived from the basic "Element" (which defaults to owl:Thing). It
>hence spares me some work if things like equivalentClass and
>equivalentProperty are at least childs of sameAs and not completely
>independent. The same is true for subClassOf and subPropertyOf.

This is actually very complicated semantically - the discussion of 
the issue is at
http://www.w3.org/2001/sw/WebOnt/webont-issues.html#I4.6-EquivalentTo

I will sum up very informally here, if you then want more info you 
can look in our semantics document or send a follow up question

Essentially, there is a difference between classes and properties vs. 
instances with respect to being the "same" -- roughly the question is 
whether when I say two classes are the same, do I mean they have the 
same elements or they designate exactly the same thing?  If I say 
"All Elephants the size of a pea" and "All mice the size of a house" 
I have two classes which have the same elements (In this case none), 
but clearly designate different things (anything in the first class 
would have to be an elephant, anything in the second class couldn't) 
-- for instances this doesn't come up.  Using the two separate 
constructs, with the slight differences in semantics, makes it 
clearer to a reasoner which is meant.  Because of this, 
owl:equivalentTo and owl:sameAs actually have different semantics - 
and thus is not the parent-child relation you want.
  This is a very informal answer to give you the intuition, please 
look at the Reference and Semantics documents for the details of the 
solution, or the issues list to see some of the discussion the WG had.


>
>5) More a beauty question. Is it necessary to still have  rdfs:XXX
>defintinions in the OWL name space? Almost all classes and properties defined
>by OWL are also rooted in the owl name space. I have just recently discovered
>that some are not (such as, I think,  rdfs:id and rdfs:subPropertyOf (?) ).
>For me it would be a bit simpler  (and cleaner) if all the basic language
>features that have to be hardcoded into the library (because the program
>needs to react to them) belonged to the OWL name space.

This is also a harder question than you might think, and one the WG 
spent a long time grappling with - see
  http://www.w3.org/2001/sw/WebOnt/webont-issues.html#I5.20-should-OWL-provide-synonyms-for-RDF-and-RDFS-objects

Roughly speaking, there is a tension between making things easier for 
the OWL user by having everything be in one namespace, or for keeping 
things defined in their approved namespaces as an architectural 
preference (i.e. by having things defined in the RDF space, the OWL 
namespace doesn't contain a lot of redundancy that would need to be 
separately maintained).  We decided that, in my own words, Web 
Architecture wins out over OWL-chauvinism :->  (The WG decision is 
linked to the issue above, so you can see the official rationale)

>
>OK, that's enough for the beginning.
>Greetings and Thanks for any response!
>
>Raik
>
>Raik Gruenberg
>Unite de Bioinformatique Structurale
>Institut Pasteur, Paris


We appreciate your questions/suggestions.  If you have questions 
about using OWL or about building tools which use OWL, the mailing 
list www-rdf-logic@w3.org is full of people who love discussing the 
how-tos and is a good place to go for answers.

  -Jim Hendler
   WOWG CoChair

-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  *** 240-277-3388 (Cell)
http://www.cs.umd.edu/users/hendler      *** NOTE CHANGED CELL NUMBER ***
Received on Wednesday, 24 September 2003 08:00:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:43:29 GMT