Re: SUO: RE: Ontology, Epistemology, Semiotics
Cathy Legg wrote:
>On Sat, 13 Sep 2003, Pierre Grenon wrote:
>
>
>
>>>>However it seems that this is true also in hogher arity claims: John is
>>>>
>>>>
>>between
>>
>>
>>>>Mary and Rosamonde according to Wilbert. It does not seem reducible to any
>>>>combination of ternary claims by the standards of the analysis of ternary
>>>>claims themselves.
>>>>
>>>>
>>>(isBetween BETWEENNESS_RELATION001 John)
>>>(delimitsBetweenness BETWEENNESS_RELATION001 Mary)
>>>(delimitsBetweenness BETWEENNESS_RELATION001 Rosamonde)
>>>(avowedBy BETWEENNESS_RELATION001 Wilbert)
>>>
>>>And then break that up into 2 triads as I showed in earlier messages....
>>>
>>>Sorry, that's the last one, I promise. :-)
>>>
>>>
>>Hey Cathy, you won#t get away so easily with this. :)
>>
>>I can't find the time now, though I promise I#ll think more about this and try
>>to see how you propose to break these following the earlier example. I guess
>>the previous example didn't completely enlight me. It still isn't clear to me
>>what#s this thridness which is preserved when you break down further. From what
>>I can tell, it's the number three in a somewhat arbitrary fashion. (Seems that
>>you can always pick up your three things and say 'see there are three of
>>them'.) In the earlier example, it's even less clear to me as equality as
>>precisely the effect of removing the putative distinction... So I guess I'm
>>still as confused as to where or what thirdness is and how it is not a trivial
>>claim which has no particular fundamental or practical significantion (but for
>>the Peirced ones?)
>>
>>
>
>Fair enough.
>
>The thirdness is just about the arity of the relations involved. BUT there
>is a complication because in a representation like the one I gave above,
>not all the logically relevant relations are explicitly stated in the form
>of predicates. There is a masked identity relation linking the tokens of
>BETWEENNESS_RELATION001, which is represented not via a predicate but
>iconically via the fact that the character-strings in each case are the
>same (which we interpret as an identity relation without realizing it,
>having been trained to do so).
>
Well, I think there's a substantial contingent of us who
imagine we see the relation of
the table/relation/RDBMS/SQL/DBMS Schema world to practical ontology
much more clearly than we see the relation of
the diagram/nodes-and-arcs/conceptual graphs/meta/metmeta world to
practical ontoloogy.
So I want to ask whether you believe that this masked-identity-relation
stuff is easily unmasked, or not.
Until I read your statement above, I had been assuming that you
were tacitly assuming something like the following.
==========
We need (but often cheat on using) something like
-Codd's RM/T "surrogates"
-Object-Oriented DBMS' object IDs ("OID"s)
-Smalltalk/Java OIDs.
-Microsoft Security ID (SID)
These surrogates tell us NOTHING about an entity
(except maybe that it has been "noticed"). (Actually the msft SID
fails, here.)
When surrogates match, they refer to the same entity.
We know how to test whether 2 surrogates match.
There are an unlimited number of surrogates, compared to the universe of
entities of interest.
A DBMS "primary key" is not suitable as a surrogate, if it means
something additional.
(E.g., A primary key which is the serial number of a vehicle is
no good as a surrogate for the vehicle, because the serial number
might get changed.)
A never-before-used and never-again-to-be used surrogate is attached to
an entity when we
begin to talk about it.
==========
So, if I take, for example, your BETWEENNESS_RELATION001 in this sense,
(as a surrogate) am I still better off using, say, a conceptual graph?
>
>This is why Peirce preferred his graphical notation over such algebraic
>notations - it is more perspicuous because it gets all the logically
>relevant relations out on the table and represents them in the same way,
>instead of the mish-mash above. Thus (in answer to an earlier message of
>Rich's) this is not a mere matter of representational ease.
>
>John Sowa has been making this point for the past few days but I wanted to
>say it in my own way as well :-)
>
>
>
>>As for the solution here, I find it quite ingenious. I wouldn't have expected
>>that you took token relations into consideration. (Am I correct that this is a
>>token? it seems to fail otherweise.) I'm not sure what to think about this
>>actually. At any rate, this is a quite courageous move. (Is that some Peirce
>>orthodoxy by the way or is it your craft?) It seems to me that there is an
>>obvious issue of infinite regress lurking there however.
>>
>>
>
>Yes it's a token. And it's not my idea but Peirce's (nicely explicated by
>Robert Burch, who is at Texas A&M, in the 90s).
>
>As you can see, the key is the ability to 'perform 'hypostatic
>abstraction' on any relation to make it itself into a relatum. The way
>Pierluigi put things, in terms of a continually extensible
>'triangulation', was actually a very good explication of this (thanks,
>Pierluigi). But in order for such triangulation to work, the identity
>relation must be transitive, and transitivity is an essentially triadic
>relation. (Thus the irreducibility of thirdness)
>
>Here we have a glimpse of why Peirce thought scholastic realism was so
>important -- not in metaphysics, but in logic. Quine, although explicitly
>disavowing metaphysical nominalism in a number of places (e.g. claiming to
>have an ontological commitment to classes), was nominalist in spriit in
>many ways in his logic, IMO, not least of which was his extensionalism.
>
>Cheers,
>Cathy.
>
>--------------------------------------------------------------------------
>Cathy Legg, Phd Cycorp, Inc.
>Ontologist 3721 Executive Center Dr., ste 100
>www.cyc.com Austin, TX 78731-1615
>
> download OpenCyc at http://www.opencyc.org
>--------------------------------------------------------------------------
>
>
>
>
>