RE: First Order Logic and Semantic Web RE: NPR, Godel, Semantic W eb

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

If a semantic service is considered a service, say 
something at the end of a SOAP call, then you can 
set the loop up as you wish.   The ontology layer 
looks straightforward.  The logic that works with 
it, the subject of a different post, has to deal 
with assertions that may or may not be provable, 
may be in conflict with other systems, etc.  We 
explored a lot of this last year on XML-Dev and 
the papers out there that dealt with these systems 
before suggest that operational issues of quality 
control over the ontology were critical to using 
it.   If the ontology is a means for the local 
system to reason over its own information (not 
a web oracle but a local advisor), then the local 
system builder will work hard to make sure it is 
consistent with the goals of the local service.

I believe the QOS is important to establishing 
authority.  We can't just use frequency or 
location of citation, or interreference, etc. 
These are too easy to game.  I think it is like 
comparing the OED to the Winn-Dixie Websters 
Dictionary.  Money gets quality but cannot 
eliminate all bias.  Therefore, one is careful 
about the transactions committed, tests the 
connections, and rates the service.  Legitimate 
authority is a different issue and has more to 
do with formal policy for decision making (whose 
shot is it to call).

The web is an amplifier.  Deal with it accordingly.

Y'all can kick Gates and MS at will but my experience 
is that they will do this as well or better than 
anyone else.  They have to sell a lot of copies of 
software to survive and even if the rental model 
is lucrative (doubtful because it works against 
the economic model of the customer even if it works 
for the vendor), they then have to thrive on the 
highest quality because switching services is 
easier (it is easier to get out of a lease than 
a mortgage).

Len 
http://www.mp3.com/LenBullard

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


-----Original Message-----
From: Joel Rees [mailto:rees@server.mediafusion.co.jp]

I guess I asked a different question from I thought. What I had in mind was
to suggest (question?) the possibility of designing SW so that any semantic
loop that formed would be subject to human judgment on its validity as long
as it exists. A required timeout mechanism would also be useful. But I don't
think it would be possible to enforce.