[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: NPR, Godel, Semantic Web

• From: "Bullard, Claude L (Len)" <clbullar@i...>
• To: Danny Ayers <danny@p...>,"Simon St.Laurent" <simonstl@s...>, xml-dev@l...
• Date: Tue, 08 May 2001 12:31:51 -0500

My question is, can the system detect that 
the answer is missing?  The ontologists told 
us that experience in building these systems 
made them rely on behavioral observation as 
the means to detect when the system returns 
the right answer.  In other words, like HTML, 
the design has many problems but it works for 
most of the cases it is designed for and we 
have extended it for others, and don't use 
it for some others.

We don't need a theoretically complete system. 
We need one that gets some useful things done. 
Fact is, no one has tried knowledge bases at this 
scale.  We can make predictions (negotiation 
and well-behaved agents are key) but we don't 
know how well this will work.

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

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


-----Original Message-----
From: Danny Ayers [mailto:danny@p...]

>It's not that such systems will return wrong answers - it's that there are
>right answers they cannot find.  How that would echo through a system or
>whether users would even notice the missing information isn't clear.

There are reasonable techniques for handling such circumstances - e.g.
timing out ;-)
I can't remember where but it has been said more than once that the web will
never be perfect, if you allow this then the problems don't arise.

I wonder if the thing you describe could also be expressed as a kind of
halting problem???>

• Prev by Date: Re: RDDL for describing fragment identifier facilities
• Next by Date: Re: RDDL for describing fragment identifier facilities
• Previous by thread: RE: NPR, Godel, Semantic Web
• Next by thread: FW: NPR, Godel, Semantic Web
• Index(es):
  • Date
  • Thread