Re: Spec doesn't talk about two-valued relationships

Graham Klyne <GK@NineByNine.org> wrote:

>>>> Because a generic system doesn't know whether 0 means false, or an address,
>>>> or whatever.
>>> It may be that we come at this with different worldviews/assumptions about
>>> how systems might work, but it seems to me that that kind of "knowledge"
>>> would be embedded in inference rules; e.g.
>>> 
>>> <http://www.aaronsw.com/> bob:chocolateLover "0" .
>>> bob:SweetBrownStuff rdf:type bob:Chocolate.
>>> ->
>>> <http://www.aaronsw.com/> bob:doesNotEat bob:SweetBrownStuff
>> The question is what rules/terms are needed to be able to do this in the
>> general case. That is, I'd like my system not to have to have specific
>> knowledge about chocolateLover, Chocolate, and doesNotEat.
> Sooner or later, methinks, it is needed that statements are grounded in
> "real-world" knowledge.  How do you suggest that such grounding may be
> introduced into a system?

I I follow what you're saying correctly, then the answer is that the
"grounding" will take place when you go to pull data out of the system.
Example: If I'm throwing a party and I need to know if you like chocolate,
that's the question I'll ask. I could really care less how that information
got there. If we use some sort of generic yes/no terminology, then my system
will be able to figure that out without any special programming.

I just want my pizza... I mean chocolate... no, wait, I don't.... oh, never
mind.

-- 
[ Aaron Swartz | me@aaronsw.com | http://www.aaronsw.com ]

Received on Sunday, 11 March 2001 15:35:28 UTC