W3C home > Mailing lists > Public > semantic-web@w3.org > March 2007

XType considered harmful

From: Richard H. McCullough <rhm@PioneerCA.com>
Date: Sun, 4 Mar 2007 18:08:46 -0800
Message-ID: <5BA3C7835B5B4EB28A6DA32D780C337D@rhmlaptop>
To: "Hans Teijgeler" <hans.teijgeler@quicknet.nl>
Cc: "Semantic Web at W3C" <semantic-web@w3.org>, "OWL at W3C" <www-rdf-logic@w3.org>, "Rob McCool" <robm@robm.com>, "Larry Lefkowitz" <larry@cyc.com>, "John De Oliveira" <johndcyc@gmail.com>

I have previously argued that you should not use both "X" and "XType"
classes in your ISO ontology.  My argument was based on my intuition
gained from working on ontologies for many years.
Now I have a concrete example, from the TAP knowledge base, showing
that using "XType" just does not make sense.

See http://mKRmKE.net/knowledge/tap/BADProductType.mkr
for the details of the mKE responses to:
    do find od *Color* in BOOT done;
    LipColor isa* ?;
    Product Type isc ?;
    ProductType isc* ?;
I was originally trying to find "Color" (a.g., red, blue, green).
Not finding any "Color", I decided to explore "LipColor".
The answer to my first question shows 5 separate paths to "Resource"!
The first path makes sense
This path is part of a "normal" ontological partition of Resource.
The other 4 paths essentially reproduce this path with one or more concepts
deleted, and the last 3 concepts replaced by
    Product Type\Class\Resource
Using both "Product" and "Product Type" has cluttered the hierarchy with
worthless information.

You can see this even more clearly in
The Resource/Class/ProductType sub hierarchy reproduces pieces of
the Resource/Tangible/Product sub hierarchy.

As I stated before, the correct approach is to let "Class" and "XType" be
sets, which are related to other classes by the "imem" relation.

Dick McCullough
mKE do enhance od "Real Intelligence" done;
knowledge := man do identify od existent done;
knowledge haspart proposition list;
Received on Monday, 5 March 2007 02:10:31 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:44:59 UTC