OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   NIH in multiple facets

[ Lists Home | Date Index | Thread Index ]

> Roger L. Costello wrote:
> > The namespace of the elements indicates the Ontology.  As I mentioned
> > earlier, an Ontology can evolve in a distributed fashion.  There need
> > not be a centralized committee of experts.  Besides, even if there was,
> > I would argue that creating a centralized committee to create a logical
> > design (i.e., an Ontology), and then allowing (encouraging!) diversity
> > of physical expressions ain't a bad way to go!
> > 
> > Naturally, you will most likely want to involve domain experts in
> > deciding what are the fundamental terms and their relationships.
> Some of the most interesting (practical) results in this area are in 
> reverse engineering ontologies and domain models from existing data 
> sources. That is, knowledge scraping seems to work as well or better 
> than any top down approach - which is both consistent with Web 
> technologies and avoids streaks of NIH, which tends to plague 
> ontology work.

I had to read this twice.  The first time I thought you meant the (in)famous 
National Institute of Health C++ class library which stubbornly imported the 
Smalltalk idea of crushing the entire world into a single  abstract class 
hierarchy.  It was the centerpiece of the debate over the epic OO debate 
between inheritance-based polymorphism and generic programming.

Then I figured perhaps you mean "Not invented here".

The funny thing is that both of these facets of "NIH" are deadly to any 
ambitions of ontologies in real life, and I've seen dangerous elements of both 
in ontological work in the field.

I think that for ontologies to work, they should be very careful about 
minimizing rigid constructs, which most often mean favoring local constraints 
over explicit interitance.  Almost everytime I run into entailment problems in 
OWL-like languages, there is concrete subClassOf lurking nearby.  Developers 
should also be very conscientious about seeking out ontologies developed by 
other parties for import, rather than reinventing the wheel.  I think this is 
much more important than worrying about whether the other ontology is *just* 
the right fit for your application.

Uche Ogbuji                                    Fourthought, Inc.
http://uche.ogbuji.net    http://4Suite.org    http://fourthought.com
Gems From the [Python/XML] Archives - http://www.xml.com/pub/a/2003/04/09/py-xm
Introducing N-Triples - http://www-106.ibm.com/developerworks/xml/library/x-thi
Use internal references in XML vocabularies - http://www-106.ibm.com/developerw
EXSLT by example - http://www-106.ibm.com/developerworks/library/x-exslt.html
The worry about program wizards - http://www.adtmag.com/article.asp?id=7238
Use rdf:about and rdf:ID effectively in RDF/XML - http://www-106.ibm.com/develo


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS