ontology-summit
[Top] [All Lists]

Re: [ontology-summit] [ReusableContent] Partitioning the problem

To: Ontology Summit 2014 discussion <ontology-summit@xxxxxxxxxxxxxxxx>
From: "Barkmeyer, Edward J" <edward.barkmeyer@xxxxxxxx>
Date: Thu, 6 Feb 2014 19:52:13 +0000
Message-id: <ea451aa793b54107ba5da82bc6571ab1@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>

John,

 

As I said, in English we have a way of providing distinct labels – “the class ‘aircraft’” vs. “aircraft”.  I would have said that the latter refers to the extension – “any thing that is an aircraft”, while the former refers only to the intension – the nature of being an aircraft.  One of my favorite examples, although for another purpose, is:

 

John Smith is a teacher.  (John Smith is an instance of ‘teacher’.)

Teacher is a profession.  (‘teacher’ is an instance of ‘profession’.)

 

The first sentence uses ‘teacher’ in the usual way.  It is about instances (extension) of the ‘teacher’ class.

The second sentence uses ‘teacher’ to refer to the ‘class’ itself, and not to its extension.  No teacher is an instance of profession.

 

There are some technical semantic issues here that I definitely don’t want to get into.  I only say that in English we do have syntax for distinguishing the intension and the extension, which may or may not be what you mean by  “an extension label different from an intension label”.

 

-Ed

 

From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John McClure
Sent: Wednesday, February 05, 2014 6:32 PM
To: Ontology Summit 2014 discussion
Subject: Re: [ontology-summit] [ReusableContent] Partitioning the problem

 

Ed and others,

When you say "RDF does not care; OWL does" that's nothing more than saying OWL added axiom support to RDF. Those axioms, taken together as a single unit, define an intension, eg the set of constrained properties applicable to each member of the extension. As a unit, an intension is associated to an extension via a relation we call "subClassOf". So the label for a class may be (dare I say) denoting an extension while at the same time, connoting an intension.

What I hear is this: the OWL flaw is that rdfs:subClassOf has unbounded cardinality, so therefore no single resource exists to which one can attach a label that denotes the intension definition, a label that may be a referent in LOD contexts separately from a label for an extension.

OK, now my question.  Can anyone cite a use case showing a need for an extension label different from an intension label? What would that label look like, something like "Class X" vs "Class X Template"?

thanks/jmc

On 2/5/2014 1:13 PM, Barkmeyer, Edward J wrote:

John,

 

We will agree to disagree.  I was using the terminology as defined in the discipline, in order to make the distinction clear. In short, ‘connotation’ and ‘denotation’ are terms-of-art in formal logic and computational linguistics and some formal semantics circles, and I do realize that they are unexpected outside those circles.  I personally recognized those definitions as ‘anti-mainstream’ when I first encountered them.  I just prefer using the terminology of the established discipline, however weird, to trying to sort out a terminological issue (reference) by inventing my own.  (In a similar way, John Sowa and I talked past one another in using the term ‘model’, which, as a term-of-art in logic, refers to a population, a given universe of discourse.)  This is a rat-hole we don’t need to go further down.

 

The important point is that the ‘referent’ of a term can be either the intension (the class, as a set of common properties)  or the extension (its members) or both (in some sense).  RDF does not care; OWL does.   The LinkedOpenData technology is founded on RDF IRIs as ‘terms’ that refer to (extensional) individuals.  In that view, memberOf and instanceOf and subclassOf are just verbs that relate to two ‘individual things’ that have IRIs.  In OWL, those verbs have well-defined semantics, and that semantics is only well-defined when the subject and object roles of the triple are played by the proper kinds of things, and OWL distinguishes ‘class’ from ‘individual’, although both are still nominally extensional.  So we are still trying to define the relationship between the structured ontological approach to capturing knowledge and the LOD approach to capturing knowledge, and to define the discipline for implementing that. 

 

Michael and Kingsley were discussing aspects of this idea, but I was confused by their wording.  So I tried to clarify, and apparently made the problem worse.

 

-Ed

 

“The best laid plans of mice and men gang aft agley.”

-- Robert Burns, “Of mice and men”

 

 

 

From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John McClure
Sent: Wednesday, February 05, 2014 3:34 PM
To: Ontology Summit 2014 discussion
Subject: Re: [ontology-summit] [ReusableContent] Partitioning the problem

 

Ed,
We may end up agreeing to disagree on this, but to me the English is abundantly clear, even in the tertiary M/W definition you cite: connation is referring to each within a plurality of things ('group of properties'), and denotation is referring to a singular thing ('the totality'). Their Latin roots indicate such too: con (with) and de (from, of) modifying note (word). Thus a name of a class denotes its extension, and connotes the members of that extension. Maybe another look at your earlier note in this light would be useful.

I'll look at the second topic later -- there's much about SBVR I need to absorb -- because reference vs denotation sounds fundamentally the basis for the claim that (paraphrasing from memory here) LOD doesn't distinguish between individuals and classes ...??? If not true, I wouldn't want that to end up in a communique and if true, it certainly SHOULD be in a communique, no?

On 2/5/2014 11:28 AM, Barkmeyer, Edward J wrote:

John,

 

1.  I did say: “In formal terminology, ...”.  Let me refer you to definition (3) of ‘connotation’ in my copy of Merriam-Webster:  “an essential property or group of properties of a thing named by a term, in logic.”  And ibidem, “denotation” (4) “the totality of things to which a term is applicable, esp. in logic”.  I admit that only a computational linguist would write ‘connote’ and ‘denote’ as the back-formations, but let me assure you, they do so, and regularly.

 

2.  Did you look carefully at the SBVR definition?  “verbal designation of a general concept in a specific subject field”.  That looks exactly like the OED definition.  But note that it underscores ‘designation’, which means that designation has the SBVR meaning, selected from possible OED meanings.  Now, SBVR clause 8.4.1 defines ‘designation’ as a ‘representation’ relationship between an _expression_ and a concept.  That definition chooses between the relationship and a role in it, which are different OED definitions:  “The designation of x by y” is the relationship, but “the designation for X” is a reference to the _expression_ in the ‘signifier’ role (per SBVR, and ISO 1087-1).  Semantically, these concepts are importantly different, because the instances of the relationship are ‘states of affairs’, while the instances of the role are ‘expressions’.  I much doubt that the OED intends ‘term’ to refer to a state of affairs.  This error (and I strongly believe it is one, precisely because SBVR makes such distinctions) arises from using a term _expression_ out of its context (and in another), so that the usage acquires a different meaning.  To be consistent with the OED intent, using definitions of the underscored terms, the SBVR definition should be:  “verbal signifier for a general concept in a specific subject field.”

 

So, I stand by what I said. 

 

It is unsafe to wade casually in these waters.  The currents are tricky.  And that is exactly how some would-be ontology reuses drown.  I come back to my previous points about reuse:  The ontology you want to reuse must be consistent with your theory, and you have to KNOW that it is consistent, not just assume that the interesting terms mean the same thing.

 

-Ed

 

P.S.  The SBVR “finalization” team formally asked the authors of ISO 1087 which of the OED definitions they meant by ‘designation’, and got two conflicting answers and one ‘no one ever asked that’.  ISO 1087 is about how to develop good vocabularies and how to write good definitions, but unlike ontologies, it acknowledges that a term may have multiple definitions that are distinguished in context.  But ‘in context’ means ‘in a usage’, not (necessarily) ‘in a namespace’.

 

 

From: ontology-summit-bounces@xxxxxxxxxxxxxxxx [mailto:ontology-summit-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John McClure
Sent: Wednesday, February 05, 2014 1:15 PM
To: ontology-summit@xxxxxxxxxxxxxxxx
Subject: Re: [ontology-summit] [ReusableContent] Partitioning the problem

 

Ed,
You have connote and denote backwards. To 'connote' is to indicate things *associated with* (per the Latin con) -- in this case, applicable to members (parts) of the extension. To 'denote' is to identify a thing specifically -- in this case, applicable to the concept itself, i.e., the extension taken as a whole.

Also, to label SBVR's 'term' definition a 'mis-definition' is odd, because it is an OED definition. I put more faith in OED definitions than apparently domain-specific definitions such as you describe.

/jmc

On 2/5/2014 9:34 AM, Barkmeyer, Edward J wrote:

Now, except for the mis-definition in SBVR (in which a 'term' is the relationship between an _expression_ and a concept), I think we can all agree that a 'term' is a linguistic _expression_ in the role of reference to a <something>.  In formal terminology, a term is said to *connote* a general concept (an intension, a predicate), which may or may not be what is meant by "class", and to *denote* its members -- the things in a given UoD that satisfy the predicate.  And in our speech, both formally and informally, we use both of those relationships.  In describing a Tbox, we are actually saying things about BOTH the nature of the intension (the connotation) and the properties of each individual in its extension (the denotation).  That is, the term 'term' is not "slippery"; <
 /o:p>
                
 it is u
 
                
nderstood to have two simultaneous notions of 'referent'.
 
              
 
              

 

 
            
 
            



              

              

              

 
            
 
 
            
_________________________________________________________________
 
            
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
 
            
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
 
            
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
 
            
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
 
            
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014  
 
            
Community Portal: http://ontolog.cim3.net/wiki/ 
 
          
 
          

 

 
        
 
      
 
      

      
      

      
 
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014  
Community Portal: http://ontolog.cim3.net/wiki/ 
 
    
 
    

  

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014  
Community Portal: http://ontolog.cim3.net/wiki/     (01)
<Prev in Thread] Current Thread [Next in Thread>