W3C home > Mailing lists > Public > public-sws-ig@w3.org > December 2003

Re: Cross-ontologies reasoning

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Mon, 15 Dec 2003 22:00:43 -0500 (EST)
Message-Id: <200312160300.hBG30hV26710@pantheon-po01.its.yale.edu>
To: public-sws-ig@w3.org


   [Francis McCabe]
      Notwithstanding the technologies being discussed, *translation*  
   between ontologies is about as tractable in the general case as mapping  
   between English and Japanese.

This assessment is overly pessimistic.  We're not talking about
translating Japanese literature into English.  In most cases the
differences between ontologies fall into categories such as these:

* One ontology represents a concept as a class, the other as a property

* One ontology makes fine distinctions about a concept; the other uses
  a broader brush.

* One ontology uses a predicate with n arguments where the other uses
  a similar predicate with n+1.  The missing argument must be deleted
  or inferred somehow.

* and so forth

Translating back and forth can be done by straightforward deductions.

Perhaps you meant merely to say that the deductions would end up
consuming exponential amounts of time.  (Which is _not_ the problem
with translating between two natural languages, such as English and
Japanese!)  You may be right, but it's not obvious.

Or maybe you meant to say that the translation rules could not be
generated automatically.  I agree with you there.

-- 
                                   -- Drew McDermott
                                      Yale Computer Science Department
Received on Monday, 15 December 2003 22:00:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 16 March 2008 00:10:53 GMT