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

Instead of <http://www.aaronsw.com/> bob:chocolateLover "0", how about
defining a class called truth and two resources of type truth called "True"
and "False" (i.e. the way the spec advocates doing enumerations)?  Then it
becomes:

<http://www.aaronsw.com/> bob:chocolateLover <:True> or 
<http://www.aaronsw.com/> bob:chocolateLover <:False>

Regards

Lee

-----Original Message-----
From: Aaron Swartz [mailto:aswartz@upclink.com]
Sent: 09 March 2001 18:41
To: Graham Klyne
Cc: RDF Interest
Subject: Re: Spec doesn't talk about two-valued relationships


Graham Klyne <GK@NineByNine.org> wrote:

>> It's true that rdf:type gets close to this,
>> but there is a general need for negation in this case, even without
getting
>> into logic and all that.
> I think that's where DAML-ONT (or whatever it's called these days) comes
in
> to play.

Perhaps we can add this to DAML -- I know they have disjoint, but I don't
think that's the same as opposite...

>> If we don't add these properties, I think we'll see
>> a ton of:
>> 
>> <http://www.aaronsw.com/> bob:chocolateLover "0" .
>> 
>> which is nowhere near as useful.
> Why less useful?   I think you can make just as many inferences from
> statements like this.  Maybe even more, because you'd be using a
> domain-specific property with possibility for more precise domain/range
> inferences.

Because a generic system doesn't know whether 0 means false, or an address,
or whatever.

-- 
Aaron Swartz <me@aaronsw.com>|           my.info
  <http://www.aaronsw.com>   |   <http://my.theinfo.org>
AIM: JediOfPi | ICQ: 33158237|  the future of news, today