[Top] [All Lists]

Re: [ontolog-forum] Guo's word senses and Foundational Ontologies

To: Azamat <abdoul@xxxxxxxxxxxxxx>
Cc: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sun, 31 May 2009 03:15:45 -0400
Message-id: <4A222EA1.4000204@xxxxxxxxxxx>
Dear Matthew and Pat,    (01)

I'll address your points first, since they are central to the main
theme.  The other notes get into issues that are also important,
but not directly related to the central focus of this thread.    (02)

MW> What (I think) Pat is proposing is to produce one ontology into
 > which others could be translated/mapped. Those other ontologies
 > need not be changed at all, so a 3D and a 4D ontology could each
 > be mapped to the "universal" ontology without having to give up
 > their own commitments.    (03)

That depends on how detailed your mapping happens to be.  If the
foundational ontology happens to have underspecified categories
with very few axioms, and you don't expect to preserve all the
information in your source ontology, then you could do some such
mapping.    (04)

But if you intended to map all the detailed axioms of a 4D ontology
into the FO in such a way that they could be fully recovered by a
reverse translation, then all those commitments would have to be
represented in the terms of the FO.  If that FO could also preserve
all the commitments of a 3D ontology, then it would have to be
able to accommodate two very different systems simultaneously.    (05)

That is a very strong constraint to lay on the FO, and I doubt
that it could lead to a consistent and successful system.  If
anyone believes that it can be done, that would make a good
research project.  But it is not yet suitable for a standard.    (06)

MW> Well, for the record, I conjecture that there is not a finite set
 > of primitives from which any and all ontologies can be defined...    (07)

PC> I am suggesting that we actually investigate the issue with a
 > proper test rather than simply assume it is impossible and
 > congratulate each other on our ignorance.    (08)

I sympathize with Matthew's conjecture.  But if Pat wants to
explore the possibility, that's his choice.  In any case, it's
a long-term research project, not something that is suitable for
a standard until some working prototypes have been implemented.    (09)

MW> I can quote examples where millions of dollars have been saved
 > in practice by taking some "ontologies" and integrating them
 > through a single ontology (ISO 15926) (whether or not it is
 > actually universal does not affect the benefits for particular
 > cases).    (010)

Your experience is consistent with Bill Andersen's reports about
their results at OntologyWorks.  They have been doing a good
business in aligning independently developed databases by means
of ontologies.  Originally, they had an upper-level ontology
based on Dolce, but as they continued, they discovered that
the upper levels were not very helpful.    (011)

Instead of an upper-level ontology, they currently use a
collection of domain-oriented ontologies that are closer to
the level of the categories in the databases they are trying
to align.    (012)

PC> I think a serviceable version suitable for testing, together
 > with some open-source utilities and example applications, can be
 > built in 3 years with a consortium of about 100 participants,
 > ca. $30M over three years.    (013)

You cannot use 100 participants unless you have something very
precisely defined for them to do.  The search for primitives is
a reasonable project for one person with a couple of assistants.    (014)

PC> OK, here is how I would do that, but I would use the Longman
 > Defining Vocabulary (LDV), since I don't actually have Guo's list.
 > Also, I think the most practical process is not to replace words
 > with LDV terms, but to use the terms one finds most appropriate,
 > and then where necessary define those with the Longman terms.
 > The *practical* defining vocabulary can expand indefinitely,
 > provided that all of the terms in that vocabulary are themselves
 > defined (recursively) by grounding in the LDV.    (015)

That sounds like a rather vague research proposal, not a plan for
using 100 people.  Until you have a very clear idea of what you
are going to implement, the efforts of at least 97 of those people
would be wasted.    (016)

You cited that paper by Yorick Wilks from 1977.  There was a lot
of interest in primitives around that time, and professors and
graduate students at several universities were exploring the
topic.  Roger Schank and his students were proposing primitives,
and some people working on conceptual schemas for database systems
were exploring related ideas.  But the lack of progress over the
past 30 years means there's no "low hanging fruit" ready to be picked.    (017)

Until someone can demonstrate (a) the existence of a suitable set
of primitives, (b) the value of primitives for defining two or three
widely different ontologies, and (c) the value of the primitives
for mapping one to the other, it is premature to fund more than
a small study project.    (018)

John    (019)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (020)

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