W3C home > Mailing lists > Public > www-webont-wg@w3.org > April 2002

Re: SUO: Re: REQUEST: survey of available ontologies, taxonomies, thesauri, lexicons?

From: Bill Andersen <andersen@ontologyworks.com>
Date: Mon, 01 Apr 2002 11:35:56 -0600
To: Charles McCathieNevile <charles@w3.org>, Leo Obrst <lobrst@mitre.org>
CC: SUO <standard-upper-ontology@ieee.org>, Ontoweb <seweb-list@cs.vu.nl>, W3C Web Ontology WG <www-webont-wg@w3.org>, RDF <www-rdf-interest@w3.org>, protege-discussion <protege-discussion@smi.stanford.edu>
Message-ID: <B8CDF69B.35E2%andersen@ontologyworks.com>
On 3/30/02 20:32, "Charles McCathieNevile" <charles@w3.org> wrote:

> To What extent would you consider an XML schema as an ontology? Some of them
> are fairly vague - XHTML, or SVG, are representing fairly broad areas of
> information without a lot of detail, but MathML and ChemML (and many other
> examples) are fairly detailed ontologies, as I understand it.

Hi, Charles...

I don't think it's really a matter of detail whether something counts as an
ontology or not.  In the weakest sense, an ontology is a collection of terms
interpreted so as to commit one (anyone who accepts the ontology) to the
existence of some class of entities.  So, in that sense the works you
mentioned (e.g. MathML and ChemML) are ontologies, albeit weak ones.

Often this commitment is implicit -- the list of terms is simply presented
and the reader/user is left to interpret it as committing to the entities
named within.  On the other end of the scale, the ontology could contain
something like a full-blown philosophically based theory of existence, which
makes it clear which entities exist and under which conditions they exist.

To lift a piece of XML-Schema from the W3C site:

  <xsd:complexType name="USAddress">
      <xsd:element name="name"   type="xsd:string"/>
      <xsd:element name="street" type="xsd:string"/>
      <xsd:element name="city"   type="xsd:string"/>
      <xsd:element name="state"  type="xsd:string"/>
      <xsd:element name="zip"    type="xsd:decimal"/>
    <xsd:attribute name="country" type="xsd:NMTOKEN" fixed="US"/>

One could take this "seriously", i.e., that all the named objects name some
real existing things (i.e., that the class USAddress is *real*) and that any
instance of it is likewise real.  But if you do, you face problems:

Here are some bad instances of USAddress (underlined):

 1)  Hauptstrasse 56, D-69118 Heidelberg, US

 2)  Ponce de Leon lived at 21 Citrus Drive, Miami FL, 12345, US

(1) is bad because Heidelberg isn't a city in the US; (2) is bad because,
while this may be a legitimate address *now*, it wasn't when Ponce was
around -- the US didn't exist -- thus this address did not exist, except as
some kind of strange possible individual.

So, how well has the schema done for us?  Not well.  And it CAN'T do better
-- it's syntax and semantics don't have enough power.   Of course one could
try to encode all of this somehow in some arcane syntax that someone is
going to have to interpret as doing what logics already do.  The RDF and
RDF-Schema efforts are just such encodings.

Note also that these remarks would be relevant no matter which side of the
ontology debate you're on; whether you believe in "what is" or "how we
talk", the criticisms I made above hold - that's why I picked 'em :-D

To do the job right, you have have to record a myriad of facts about
USAddress that the folks who wrote the above probably never even thought of
because they're thinking of electronic interchange of data between players
who's primary motivation is to get something, anything, working NOW.  This
is no condemnation.  There are real people out there with real problems who
aren't going to wait for SUO or anyone else to solve them if the WWW
community or W3C comes forward with something to help them NOW.

That said, WE SHOULD NOT fool ourselves that any of these quick fix tools
(RDF(/S), XML(/S), DAML) are going to solve the real problems of (applied)
ontology, let alone the two relatively easy ones I mentioned above.  The
long range solution to problems of data and information system integration
depend on their solution.

In this sense the artifacts you mentioned are NOT examples of Ontology, but
can be called "ontologies" if it makes us feel good to do that.


Bill Andersen
Chief Scientist, Ontology Works
1132 Annapolis Road, Suite 104
Odenton, Maryland, 21113
Mobile: 443-858-6444
Office: 410-674-7600
Web: http://www.ontologyworks.com
Received on Monday, 1 April 2002 12:36:01 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:56:42 UTC