RE: Antwort: RE: Semantic web article in Nature Biotechnology

- Phillip, 

> So, in other words, the relationship "foo" will mean 
> different things in different ontologies, depending on which 
> profile they decide to use? 
 
Yes or no.  The meaning of the 'foo' never change. When you write your
ontology, you explicit its meaning.  But you cannot prevent others to
interprete in their own way.  Besides, isn't it a reality that we might have
different conceptualization on the same word?  The profile is an application
logic, a personal viewpoint, or a context. The evolve in parrel with
ontology.  Eventually someone popular Profile will be the most predominant
one.  
 
> If it turns out to be faulty, then they can all be fixed at once. 
> Is this not one of the main purposes of re-use?

No. (1) Can you guarantee that the change in core will not cause any
inconsistency on the dependent ontology?  If the answer is not, your claim
don't stand. (2) How about the application, you may kill all applications
that dependent on the prior version.  Profile helps you with a graceful
evolution.

> It's a modularisation problem. If we use a common relationship 
> ontology, and I import your ontology into mine to use, for example,
> a few of the terms, then, potentially, any of your terms can appears 
> as a child of mine.  

If you use one term in an ontology, you import the entire ontology.
Ontology is the modulation concepts.  
 
> I'm not really sure. Standards for developing ontologies would 
> seem to have major implications for deployment to me. 

There won't be any hard and objective criteria to partition knowledge, only
subjective one.  But there are physical means to distribute the knowledge,
like in books, articles, web page etc.  I don't think you can develop a
"standard" on the former because you cannot licence the content of a book,
but you probably can develop standard for the later on how the content is
boundled physically.  Developing ontology is like to write a book and
deploying ontology is like to print a book.  Profile can help later but not
the former, which has some principles that are helpful but I will leave it
at other time.

Xiaoshu

Received on Thursday, 13 October 2005 14:25:50 UTC