W3C home > Mailing lists > Public > semantic-web@w3.org > August 2008

Re: Thing and Class

From: Harry Halpin <hhalpin@ibiblio.org>
Date: Thu, 28 Aug 2008 02:19:58 +0100
Message-ID: <48B5FD3E.8080702@ibiblio.org>
To: "Richard H. McCullough" <rhm@pioneerca.com>
Cc: Jack Krupansky <jack@basetechnology.com>, Semantic Web at W3C <semantic-web@w3.org>

Richard H. McCullough wrote:
> I'm first trying to understand the definitions of "Class" and "Thing"
> in RDF/OWL.
>
> In the past, I have suggested different definitions, which I would have
> used in mKR.  But I only did that because I couldn't understand the
> RDF/OWL definitions.  I'm still trying to understand.
>
> I'm looking for help, for clarity.
>
> I think these questions are appropriate topics for the Semantic Web.
> If these questions can not be answered, then your "Semantic" Web
> is manipulating meaningless symbols.
They are appropriate assuming you have looked at the specs and are 
asking a reasonable question, not "Hey, I happen to have my own 
idiosyncratic way of using these words, why doesn't the SemWeb agree, 
and by the way, I haven't looked at the specs."
 
Here's the specs. Note that OWL is not doing metaphysics per se, but 
it's a knowledge representation language that like every KR language, 
including supposedly "perfect" ones, makes some design choices. You may 
not agree with them, that's fine, you can use another one or create your 
own (which you seem to have done).

However, I think the spec writers were reasonably and admirably clear, 
providing both an English definition and a formal one (that's about as 
clear as you get once you get your head around the formalism). While 
their definitions may be different than your 'common-sense' ones, 
remember that people have radically different intuitions about the 
meaning of 'thing' metaphysically. I for one endorse the stance of 
Alfred Whitehead, while you like Ayn Rand. That's fine - the SemWeb OWL 
WG made its own choices here, and one should not, read any "magic" - 
much less "metaphyiscal assumptions"  into the mnemonics used by OWL 
just because certain English words were used in their URIs. Remember URI 
Opacity!
 
So, back to the specs:

English Language Definition of OWL Class:

"Classes provide an abstraction mechanism for grouping resources with 
similar characteristics. Like RDF classes, every OWL class is associated 
with a set of individuals, called the /class extension/. The individuals 
in the class extension are called the /instances/ of the class. A class 
has an intensional meaning (the underlying concept) which is related but 
not equal to its class extension. Thus, two classes may have the same 
class extension, but still be different classes."[1]

English Language Definition of owl:Thing:
You should probably read: "Two OWL class identifiers are predefined, 
namely the classes  owl:Thing and  owl:Nothing. The class extension of 
owl:Thing is the set of all individuals. The class extension of 
owl:Nothing is the empty set. Consequently, every OWL class is a 
subclass of owl:Thing and owl:Nothing is a subclass of every class" [2]

I think the problem you are encountering is that owl:Thing is used to 
introduce individuals often, since all individuals are a subclass of 
owl:Thing, one by default if one wishes to say very little about a new 
individual one can simply state it's a member of the class of owl:Thing, 
since *every* individual belongs to owl:Thing (except owl:Nothing and in 
OWL-DL, datatyped literals). If one wishes to say more, one can use a 
new class, which identifies a subset of owl:Thing and so is a sub-class 
of owl:Thing.

Formal Definition of owl:Thing
Note that since OWL-DL tends to divide resources into individuals (ABox) 
and classes (TBox) and RDF allows these resources to be mixed up, 
there's two formal definitions, one for OWL DL/Lite and another 
compatible with RDF.

OWL:
http://www.w3.org/TR/owl-semantics/direct.html#owl_Class_semantics
RDF:
http://www.w3.org/TR/owl-semantics/rdfs.html#owl_Class_rdf


OWL:
http://www.w3.org/TR/owl-semantics/direct.html#owl_Thing_semantics
RDF:
http://www.w3.org/TR/owl-semantics/rdfs.html#owl_Thing_rdf

If you find this all indecipherable, I believe Hendler and others have 
put out pretty good books on the subject you may find of interest. 
Unforunately there isn't yet a good text I have found on "How to Read 
Formal Semantics" but maybe someone else on the list has recommendations.

I am afraid I won't have time to answer any more hopefully reasonable 
questions on this matter, but maybe some of the OWL experts on the list 
can help with any more questions. I'd phrase the questions in OWL 
though, rather than a custom KR language that others probably aren't 
familiar with or don't use. If examples are in MKE or whatever, please 
keep that *off* SemWeb lists, as it's isn't part of the SemWeb and  
trust me, if you think you are having a hard time understanding the 
SemWeb, most people have on the SemWeb list can't make heads or tails of 
an MKE example.

[1] http://www.w3.org/TR/owl-ref/#Class
[2] http://www.w3.org/TR/owl-ref/#ClassDescription
> Dick McCullough
> Ayn Rand do speak od mKR done;
> mKE do enhance od Real Intelligence done;
> knowledge := man do identify od existent done;
> knowledge haspart proposition list;
> http://mKRmKE.org/
>
> ----- Original Message ----- From: "Harry Halpin" <hhalpin@ibiblio.org>
> To: "Richard H. McCullough" <rhm@pioneerca.com>
> Cc: "Jack Krupansky" <jack@basetechnology.com>; "Semantic Web at W3C" 
> <semantic-web@w3.org>
> Sent: Wednesday, August 27, 2008 5:01 PM
> Subject: Re: Thing and Class
>
>
>> While I'm sure we all appreciate the work done on MKR, could we keep 
>> MKR-specific posts to a MKR-list, not a Semantic Web list. While 
>> Richard (Dick) McCullough is free to develop any interpretation of 
>> "class" and "thing" he wants in MKR, of course, but the Semantic Web 
>> community uses the ones in the W3C Specs, and I thought they were 
>> clear in the specs. If MKR disagrees, that's fine (there's a long 
>> tradition of disagreeing about rather vague high-level concepts like 
>> 'class' and 'thing'), but that's not a question about the  Semantic Web.
>>
>> Richard H. McCullough wrote:
>>>
>>> I want to banish Class to the bookkeeping context, where it belongs.
>>>
>>> When two classes are equivalent, it means they have the same members,
>>> but different definitions -- which means they are in different 
>>> contexts.
>>>
>>> When you merge the two contexts together, you get confusion.
>>> Which class are we talking about now -- Class or Thing?
>>> Which definition are we talking about now -- Class or Thing?
>>>
>>> The class "cup" abstracts all properties of its individual member 
>>> "cups".
>>> That includes how a cup is used, what a cup is made of, etc.
>>>    cup subClassOf Thing;
>>> includes all those properties.  We might call this cup-the-Thing
>>> But when you say
>>>    cup type Class;
>>> you're in a different context - talking about cup-the-Class.
>>> That's what I refer to as the bookkeeping context.
>>>
>>> If you insist on dragging Class into the Thing context,
>>> then I recommend doing it in the form of a ClassSet.
>>>    cup ismem ClassSet;
>>>    ClassSet type Set;
>>>    Set  subClassOf  Thing;
>>>
>>> Dick McCullough
>>> Ayn Rand do speak od mKR done;
>>> mKE do enhance od Real Intelligence done;
>>> knowledge := man do identify od existent done;
>>> knowledge haspart proposition list;
>>> http://mKRmKE.org/
>>>
>>> ----- Original Message ----- From: "Jack Krupansky" 
>>> <jack@basetechnology.com>
>>> To: "Richard H. McCullough" <rhm@pioneerca.com>
>>> Sent: Wednesday, August 27, 2008 2:49 PM
>>> Subject: Re: Thing and Class
>>>
>>>
>>>> But... if you get rid of Class, doesn't it then follow that 
>>>> subClass is no longer defined or of comparable meaning?
>>>>
>>>> Is there a subThing that is "class-like"? I would imagine that 
>>>> subThing is a decomposition of a Thing into the subThing's of which 
>>>> it is composed, which is not "class-like" categorization, although 
>>>> it has some reductionist appeal. But, a purely reductionist 
>>>> analysis does not look outwards to levels of abstraction for how a 
>>>> Thing is externally viewed, perceived, and used. Two "cups" would 
>>>> not have Class "cup" that recognizes an abstraction about how a cup 
>>>> is used, but would be classified as to their material and form of 
>>>> construction as Thing's.
>>>>
>>>> -- Jack Krupansky
>>>>
>>>>
>>>> ----- Original Message ----- From: "Richard H. McCullough" 
>>>> <rhm@pioneerca.com>
>>>> To: "Semantic Web at W3C" <semantic-web@w3.org>; "KR-language" 
>>>> <KR-language@YahooGroups.com>; "cyclify austin" 
>>>> <cyclify-austin@yahoogroups.com>
>>>> Sent: Wednesday, August 27, 2008 4:59 PM
>>>> Subject: Thing and Class
>>>>
>>>>
>>>>>
>>>>> Since  Thing  subClassOf  Class;
>>>>> and     Class  subClassOf  Thing;
>>>>> it follows that  Thing  equivalentClass  Class;
>>>>>
>>>>> So, I say:  get rid of Class!
>>>>>
>>>>> Dick McCullough
>>>>> Ayn Rand do speak od mKR done;
>>>>> mKE do enhance od Real Intelligence done;
>>>>> knowledge := man do identify od existent done;
>>>>> knowledge haspart proposition list;
>>>>> http://mKRmKE.org/
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
Received on Thursday, 28 August 2008 01:20:35 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:24 GMT