[Top] [All Lists]

Re: [ontolog-forum] Defining Concept

To: Mike Brenner <mikeb@xxxxxxxxx>
Cc: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Pat Hayes <phayes@xxxxxxx>
Date: Sun, 21 Jan 2007 11:15:52 -0600
Message-id: <p06230902c1d948ad99ad@[]>
>Pat Hayes wrote:
>>  ... a universal, non-divisible idea ...
>>  Wha?? First, what distinguishes "universal" ideas
>>  from (I presume) non-universal ones? Second, what
>>  does it mean for an idea to be "divisible"?
>>  Third, what kind of combination are we talking
>>  about? I would suggest (following Fodor) that the
>>  key idea here is not the things you are calling
>>  'concepts', which in almost every detailed
>>  account that has ever been put forward turn out
>>  to be little more than nodes in a graph or points
>>  in a space, but rather the 'combinations' that
>>  they take part in, or rather still the
>>  *structure* of these combinations. It is the
>>  space (or maybe, the network, or the relational
>>  structure, or whatever one wants to call it)
>>  which matters and which gives the
>>  points/nodes/names/identifiers in it the
>>  "conceptual" structure that they have.
>I would like to answer your three questions
>in reverse order.
>The kind that our tools do not currently handle,
>which I will symbolize by "phrases".    (01)

Well, that hardly tells me what you mean. Take as 
a concrete example a full first-order 
theorem-prover as a 'tool'. This can handle 
(depending on the richness of the ontology that 
it is asked to draw inferences from) quite 
elaborate kinds of "combination". I am not sure 
what you mean to refer to by a "kind" of 
combination that cannot be 'handled' by such a 
tool. Do you mean to refer to the sheer 
computational complexity of some ontologies, 
which might defeat current SOA in machine 
inference? Or do you mean that first-order 
reasoning is inherently insufficient to the task, 
and that we need, say. probabilistic inference or 
higher-order reasoning, or some such?    (02)

>The combination (of concepts)
>ultimately refers to things like phrases,
>denoted by combinations of references
>within ontologies, with respect to their
>contexts.    (03)

I have no idea what you are talking about. Can 
you give an example? By the way, you really MUST 
say what sense of "context" you mean. There are 
literally dozens of possible meanings for this 
word.    (04)

>  I agree that words alone have
>insufficient meaning (because they
>have insufficient relationship
>with their contexts).
>Divisible means that the combined
>structures (call them hyper-ontologies,
>or, simply phrase-ontologies) have
>certain predictable circumstances in
>which the phrase can be distributed out
>(factored) into smaller phrases or words,
>or in which the context can be factored out.    (05)

Again, I have no idea what you are talking about. 
What "combined structures"? What do you mean by 
"distribution" and "factoring"?    (06)

Look, ontology engineering is by now a fairly 
hard discipline. It has long since passed the 
stage of vaguely expressed science-fictiony 
ambitions. It has a technical vocabulary rooted 
in 30-odd years experience of automated 
inference, AI Krep technology, etc., which in 
turn is based on logical ideas going back at 
least 60 years now. None of what you say above 
seems to have any relationship to the terminology 
used in this field. Can you explicate your 
thoughts using a shared technical vocabulary?    (07)

>Sometimes that factoring will result
>in retaining almost all the meaning,
>and sometimes you can factor the
>ontologies across the phrase only
>for the purpose of sub-lookups,
>sub-processing, and partially
>relating the smaller phrases to
>portions of the context or
>to somewhat related concepts.
>This does not seem to derive or
>to project from Fodor's
>generative grammars.    (08)

I wasn't referring to generative grammars, but 
rather to his monograph "Concepts: where 
cognitive science went wrong".(Oxford 1998), see 
an excellent review:
http://online.sfsu.edu/~kbach/Fodorreview.htm    (09)

>  Rather it
>comes from Wittgenstein's
>theory of levels of quotation
>Wittgenstein actually warns against
>assigning any meaning to individual
>words or names, and further warns
>against the many problems of
>making causal inferences from
>partially or unrelated facts.
>Universal infers universal applicability    (010)

Applicability of what to what? Really, none of 
this helps in the slightest to explicate what is 
being said here.    (011)

>by means of uniqueness of designation
>(the problem of naming things uniquely).    (012)

Well, it seems to me that there is no such 
"problem" since it is obviously impossible to 
give unique names to everything. We should simply 
give up on this ambition before even attempting 
it. In any case, this hardly applies to 
*concepts*.    (013)

>A system of uniquely accessible keys
>gives a database universality.
>Someday, a system of uniquely accessible
>data structures (of which ontologies
>will be one "face" or "projection"),
>will allow us to reach deep into the
>complexities of phrases    (014)

(Here, do you use this word in its ordinary 
English sense, or in the technical (and to me 
still completely obscure) sense that you indicate 
at the beginning of this email?)    (015)

>  and their
>multidimensional relationships with
>their contexts.    (016)

I have no idea what you are talking about here.    (017)

All of this seems like a rather blurry vision of 
a golden future rather like that described in 
Ursula LeGuin's "EarthSea" novels, in which 
everything has a single, unique True Name, and to 
know the True Name of a thing is to give one 
magical powers over it.    (018)

But in any case, it doesn't seem to have anything 
to do with ontologies or ontology engineering.    (019)

Pat Hayes    (020)

>The "heads" and "tails"
>of the arrows in these hyper-ontologies
>will serve as the atoms of "identity"
>that make them universal.
>Mike Brenner
>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
>    (021)

IHMC            (850)434 8903 or (650)494 3973   home
40 South Alcaniz St.    (850)202 4416   office
Pensacola                       (850)202 4440   fax
FL 32502                        (850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes    (022)

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    (023)

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