xml-dev - RE: Ontologies

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

I see your point, but there are issues where 
the tests should be performed by a credentialed 
organization.  One user of a node might not care 
about the credentials of the node, another would. 
It isn't usually a good idea to alpha test 
in the field.  As stated in an earlier email 
about the semantic web, "don't fly the first one."  
The other issues of operational credentials have 
to be dealt with differently than testing the 
node by black box means.   Would you agree that 
black box testing is a statistical test?

As Martin and Uche point out, the multilingual 
issues are difficult.  These correlate loosely 
to multi-cultural issues and these are even more 
difficult.  On the other hand, I can't envision 
ontological services without a control layer 
being sensible.  The locus of control deserves 
attention.

The contraint axioms deserve attention.  For 
example, it isn't enough to establish a unit 
of time served (for presidents or jailed 
individuals) since events control unit values. 
(a president can be impeached, die in office, 
etc., a prisoner can attempt an escape and 
get an extended sentence, and so forth).

The McGuiness article has some good insights 
into other issues of quality such as carefully 
eliminating cycles.   We can establish quality 
credentials, it seems, although we cannot eliminate 
the problem of the source of the opinion, that is, 
the legitimacy of authority in ontological commitment. 

We can test and observe.  I contend that 
the services paradigm is better for the third 
and fourth questions.

3.  By what tests (observable behaviors) 
    do we measure commitment?
4.  By what means do we initiate or 
    terminate such commitment?

Again, the lesson of the golem was the 
problem of it asserting a role as master 
instead of servant and the problem of 
turning it off once it had done this 
and caused harm to the community.  
Ontological services will be required 
to establish a contract for the services 
and both users of services and service 
providers assume a measure of culpability 
for fair use.  The struggle to establish 
controls is already underway.  

http://news.excite.com/news/ap/001219/14/library-filters

Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h


-----Original Message-----
From: W. E. Perry [mailto:wperry@fiduciary.com]

"Bullard, Claude L (Len)" wrote:

> I believe that is "black box" testing.

Testing?! Sure, I guess we can test as a black box every node whose
expertise we
might want to incorporate into a larger construct, but this is identical to
the
process of *using* that node.