ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] [Obo-relations] Heterarchy & Hierarchy, oh my my

To: Pat Hayes <phayes@xxxxxxx>
Cc: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Bill Andersen <andersen@xxxxxxxxxxxxxxxxx>
Date: Mon, 5 May 2008 10:22:11 -0400
Message-id: <FE3DE1EC-A79E-425D-9E99-F21B1D559229@xxxxxxxxxxxxxxxxx>
G'morning, Pat.    (01)

On May 5, 2008, at 09:53 , Pat Hayes wrote:    (02)

> At 8:25 AM -0400 5/5/08, Bill Andersen wrote:
>> Hi John.
>>
>> On May 4, 2008, at 23:06 , John F. Sowa wrote:
>>
>>> The useful distinction can be stated in other terms, such
>>> as types and instances.  Or even predicates and instances.
>>
>> That's not the same distinction.  You're speaking only of roles wrt  
>> an
>> instantiation relation
>>
>> (x,y) (instance(x) & type(y)) <-> instantiates(x,y)
>>
>> I don't find that very useful at all.  The distinction I'm talking
>> about is between two classes
>>
>> (x) (Type(x) <-> ~Indiv(x))
>>
>>      ... (if you don't like the word 'Universal')
>
> Apparently that isnt what OBO means by 'universal', since (for  
> example) MarriedMan isn't a universal, and for sure it isn't an  
> Indiv, right? Which to my mind only gives extra force to John's  
> point: people are using this word to mean all kinds of different  
> things. So its doing more harm than good.    (03)

Right.  I'm not worrying about OBO here and I volunteered to drop the  
name 'universal' precisely because I'm not offering any kind of theory  
of universals as they're construed in philosophy.    (04)

>>> My complaint about the terms 'universal' and 'particular'
>>> can be seen in the enormous amount of ink that has been
>>> spilled over the past few millennia on those terms.
>>>
>>> I'll agree that some of that discussion has made useful
>>> points, but any pair of terms that requires so much
>>> discussion to explain is not a pair that is pedagogically
>>> useful to perpetuate.
>>
>> To be clear, you're talking about the *names*, right?  Because even  
>> if
>> we agree not to use them, I'm going to find an ontology where I can  
>> say
>>
>>      (instantiates MadisonSquareGarden JohnSowa)
>>
>> , with these names conventionally interpreted, pretty damned hard to
>> make sense of.
>
> Well, OK, but how about
>
> (instantiates British PatHayes) &
> (instantiates CitizenshipProperty British) &
> (instantiates LegalClassification CitizenshipProperty)    (05)

Well, you swapped the argument order of what I had intended (namely  
(instantiates <thing> <class>)), but ok.    (06)

> This kind of thing is also ruled out by your strict binary  
> exclusivity rule.    (07)

No, it's not.  In your example, British, CitizenshipProperty, and  
LegalClassification would all be what I called 'Type's above, whereas  
PatHayes would not - it would be an Individual.    (08)

> And getting us past this kind of rigidity is one of the chief  
> motivations for OWL 2 replacing OWL 1.    (09)

Sorry, I don't track the OWL world closely enough.  Didn't know there  
was a ...2 in the works.  Do you mean the punning capability that OWL  
1.1 has?    (010)

> If you have managed to avoid the need for this, you must have either  
> been looking at very simple ontologies, or have some 'work-around',  
> ie an ugly hack to avoid naive binary exclusion rules.
>
>
> By the way, writing (instantiates A B) instead of (A B) is a good  
> example itself of one of these ugly hacks.    (011)

(A B) is cool with me.  I was just trying to make explicit what the  
predication was doing.    (012)

>> > My recommendation is to apply the refrigerator principle
>>> to those two terms.
>>
>> I'm happy to toss any overly burdened terms so long as the same work
>> gets done.
>
> And what work is that, in this case?    (013)

That it makes no sense to say (PatHayes ...) where we take 'PatHayes'  
to denote you.  If you can tell me how that makes any sense while  
preserving what we mean when we talk of 'Type', I'm all ears.  All  
else you said    (014)

(British PatHayes)
(CitizenshipProperty British)
(LegalClassification CitizenshipProperty)    (015)

is just fine (not just fine, but great) with me.  Nothing I said in my  
original post implied that it was not.  I think I made that pretty  
clear (and ... one of the main reasons I was involved in CL was  
because I *want* precisely what you described).    (016)

        .bill    (017)



_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (018)

<Prev in Thread] Current Thread [Next in Thread>