W3C home > Mailing lists > Public > public-sw-meaning@w3.org > April 2004

RE: How does RDF/OWL formalism relate to meanings?

From: John Black <JohnBlack@deltek.com>
Date: Fri, 9 Apr 2004 18:24:04 -0400
Message-ID: <D3C8F903E7CC024C9DA6D900A60725D90539381A@DLTKVMX1.ads.deltek.com>
To: "Pat Hayes" <phayes@ihmc.us>
Cc: <public-sw-meaning@w3.org>

> From: Pat Hayes [mailto:phayes@ihmc.us]
> Sent: Friday, April 09, 2004 5:32 PM

> >>  Let me tell you a true story about words (friends of mine stop
> >>  reading at this point as you have heard it before). Once 
> I was with 2
> >>  other folk trying to make a list of things that are found in a
> >>  typical office, so we started listing the things in the office in
> >>  which we were: desk, chairs, piles of paper, computer, 
> carpet... no.
> >>  One objected that the carpet was not in the office. Amazed, the
> >>  others tried to tell her she was wrong, eg by arguing 
> where else was
> >>  it but in the office? And she insisted that the carpet 
> was not IN the
> >>  office, but PART OF the office. There followed a lengthy (c. 1.5
> >>  hour) discussion of 'being in an office' , which ranged over such
> >>  issues as, if the door is not in the office when closed, 
> is it in the
> >>  office when it is open and projecting into the internal 
> space of the
> >>  office? (yes)...and what if you bring a can of paint into 
> the office
> >>  (2 yes), dip a brush into it and apply the paint to the 
> wall, is the
> >>  paint on the wall in the office? (1 yes, 1 no) What about when the
> >>  paint has dried, is it then in the office? (2 no, after 
> negotiation)
> >>  And so on. Eventually it gradually became clear that 
> these two adult
> >>  human native English speakers had two different senses of 
> the meaning
> >>  of words like 'office' and 'room': roughly, for one it was an
> >>  architectural abstraction, for the other something like a 
> decorated
> >>  enclosed inhabitable space with movable objects in it. 
> Both of them
> >>  were amazed and slightly horrified at the awfulness of the other's
> >>  viewpoint. And both had grown to adulthood as native speakers of
> >>  English, and were friendly colleagues,  *without ever even being
> >>  aware* that any other native speaker could hold such a weird and
> >>  alien view of what a word like "office" meant. The difference in
> >>  concepts, although quite extreme when brought into the 
> open, and easy
> >  > to reveal by statements in English - it didnt require 
> EEG traces or
> >>  surgery - had never before impinged on their everyday use 
> of English
> >>  to communicate in everyday life.
> >>
> >>  I do not think that this is atypical. If you think about 
> how language
> >>  is learned, it would be incredible if any two human beings had
> >>  *exactly* the same concepts attached to a given word. 
> Your notion of
> >>  'strength' isn't going to be exactly the same as my notion of
> >>  strength even if we devoted the rest of our lives to getting them
> >>  aligned as closely as we could.  And of course THIS DOES 
> NOT MATTER
> >>  to any practical issue in language use.  We don't need to have
> >>  perfect conceptual alignment in order to communicate successfully.
> >>
> >>  So if we can't do it for English words, why are you 
> getting into such
> >>  a tizzy about the fact that we can't do it for URIrefs ?
> >>
> >
> >Perhaps the reason I'm getting into such a "tizzy", is that I see it
> >from the opposite direction.  If it is so hard for humans, who have
> >enormously powerfull language abilities built into their brains from
> >birth, to do it without errors, then surely sw-agents will produce
> >vastly more errors when trying to do anything that even resembles
> >semantic reasoning.
> 
> But the real moral of the story is that it wasn't an error. The 
> brains of the guys in the office were working fine: the moral is that 
> there is something wrong with the meta-theory of communication that 
> says that it works by accurately conveying exactly defined 
> 'meanings'.  Wrong.  People don't NEED to have their concepts aligned 
> exactly. SW agents don't even need to have their concepts aligned at 
> all, they just have to point to the same ontologies and to keep their 
> formal semantics aligned, which ought to happen by virtue of their 
> designers reading the SW specs.

Its a great story, and I promise I'll hit those specs hard this weekend.
But I can't resist this:  The real moral of the story might change again 
if the two entered into a contract where one offered to sell his office 
contents to the other and came back to find his carpet torn out.

John


> 
> Pat
> 
> >
> >John
> >
> >
> >>
> >>  Pat
> >>  --
> >>  
> ---------------------------------------------------------------------
> >>  IHMC	(850)434 8903 or (650)494 3973   home
> >>  40 South Alcaniz St.	(850)202 4416   office
> >  > Pensacola			(850)202 4440   fax
> >>  FL 32501			(850)291 0667    cell
> >>  phayes@ihmc.us       http://www.ihmc.us/users/phayes
> >>
> >>
> 
> 
> -- 
> ---------------------------------------------------------------------
> IHMC	(850)434 8903 or (650)494 3973   home
> 40 South Alcaniz St.	(850)202 4416   office
> Pensacola			(850)202 4440   fax
> FL 32501			(850)291 0667    cell
> phayes@ihmc.us       http://www.ihmc.us/users/phayes
> 
> 
Received on Friday, 9 April 2004 18:24:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:42:16 GMT