W3C home > Mailing lists > Public > semantic-web@w3.org > May 2009

Re: AW: [Welcoming feedback] Semantic Web: Information wants to be useful

From: Pat Hayes <phayes@ihmc.us>
Date: Wed, 13 May 2009 12:22:26 -0500
Cc: "David Baxter" <retxabd@gmail.com>, <semantic-web@w3.org>
Message-Id: <14E7A112-70C8-40EF-BB72-67C9FF2F756F@ihmc.us>
To: "Paulheim, Heiko" <heiko.paulheim@sap.com>

On May 13, 2009, at 1:18 AM, Paulheim, Heiko wrote:

> Hi Pat,
>
> >PS. What does Cyc do about elements which only exist  
> macroscopically as compounds? If there are no pieces of pure  
> Ytterium, say, then the class Ytterium is the empty class. If there  
> are no pure samples of Einsteinium, similarly. In an OWL reasoner,  
> you could infer that they are the same class, hence sameAs one  
> another (since they are both classes).
>
> I feel like we should be more careful about the term "the same  
> class". Two classes are not the same only because they are both empty

Well, that depends on what ontological framework you use. They are the  
same in OWL-DL, or any ontology written in it (and intended to be used  
by OWL-DL reasoners), which is why I referred to an OWL reasoner. I  
should have said, an OWL-DL reasoner, but virtually all of them are in  
any case.

As a philosophical position, I agree with you and Cyc (and ISO Common  
Logic and RDFS and OWL-Full, I might add) , but I fear that we are in  
the minority if one were to simply count ontologies.

> - centaurs and unicorns are not the same.
>
> "NOTE: The use of owl:equivalentClass does not imply class equality.  
> Class equality means that the classes have the same intensional  
> meaning (denote the same concept). " [1]

Yes, the general overview document does say this, because it is  
designed to be read so as to cover all the OWL options, including OWL  
Full, where the class/extension distinction is maintained. However, if  
you read more deeply, you will discover that in the Semantics and  
Abstract Syntax document, which the overall spec says is "is the final  
and formally stated normative definition of the language", the  
semantic conditions on equivalentClass in OWL DL are simply that the  
classes are equal. (http://www.w3.org/TR/owl-semantics/direct.html#3.3).

But in any case, the issue we began with was the use of owl:sameAs,  
and that means identity in any version of OWL.

Pat



>
> Best,
> Heiko.
>
> [1] http://www.w3.org/TR/owl-ref/#equivalentClass-def
>
> Von: semantic-web-request@w3.org [mailto:semantic-web- 
> request@w3.org] Im Auftrag von Pat Hayes
> Gesendet: Dienstag, 12. Mai 2009 16:48
> An: David Baxter
> Cc: semantic-web@w3.org
> Betreff: Re: [Welcoming feedback] Semantic Web: Information wants to  
> be useful
>
>
> On May 12, 2009, at 9:06 AM, David Baxter wrote:
>
>> Pat Hayes said:
>>
>> > I know both Cyc and dbPedia say their
>> > concepts are sameAs one another, but they are both wrong. Cyc  
>> defines
>> > a 'piece' of carbon; dbpedia defines the chemical element carbon.
>> > These concepts are NOT owl:sameAs one another, no matter what the
>> > websites say.
>>
>>
>> Hi Pat,
>>
>> We're definitely interested in improving the quality of our  
>> owl:sameAs links to DBpedia and other datasets. In this case,  
>> however, I believe the owl:sameAs link is good -- it's the OpenCyc  
>> comment that's bad. The URI opencyc:Carbon denotes an owl:Class  
>> representing the element carbon. Its instances are individual  
>> pieces of carbon, including diamonds and lumps of coal. We'll get  
>> the comment fixed.
>
> Hmm. So - just to see if I follow you here - Cyc thinks that a  
> chemical element *is* the class of all its macroscopic pieces? Is  
> this what DBpedia also thinks a chemical element is? Because (a)  
> that seems to me to be a very idiosyncratic view of what a chemical  
> element is, and (b) if DBpedia has some other ontology of chemical- 
> element-hood, then your two concepts are very unlikely to the sameAs  
> one another. Bear in mind that owl:sameAs really does mean logically  
> identity, so ANYTHING said using one name is true using the other.  
> So someone should be able to take any DBpedia content mentioning  
> carbon, and any piece of Cyc content mentioning carbon, substitute  
> one carbon name for the other throughout both chunks, and conjoin  
> them, and the result ought to make sense in both systems. Is that  
> indeed true, in this case?
>
> Pat
>
> PS. What does Cyc do about elements which only exist macroscopically  
> as compounds? If there are no pieces of pure Ytterium, say, then the  
> class Ytterium is the empty class. If there are no pure samples of  
> Einsteinium, similarly. In an OWL reasoner, you could infer that  
> they are the same class, hence sameAs one another (since they are  
> both classes).
>
>>
>> David Baxter
>> Cycorp
>
> ------------------------------------------------------------
> IHMC                                     (850)434 8903 or (650)494  
> 3973
> 40 South Alcaniz St.           (850)202 4416   office
> Pensacola                            (850)202 4440   fax
> FL 32502                              (850)291 0667   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Wednesday, 13 May 2009 17:23:09 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:29 GMT