Re: Domain of :key

On 4/3/13 11:58 AM, Henry Story wrote:
>
>> 3) That you are unable to distinguish agents does not imply they are 
>> the same
>
> Nobody denies that.
>
> The role of cert:key and cert:identity, is an affirmation that the 
> agent the
> agent that can prove he has the private key of the given public key is 
> the agent identified
> by the given WebID.
>
> That's the game we have been playing since the beginning of WebID.

Yes.

Melvin:

We write code that implements IFP relation semantics across a plethora 
of programming language, runtime, and operating system combinations, or 
we can do by simply packing the logic into the data. RDF (which is all 
about putting semantics into the data) enables us to avert the 
aforementioned overhead via machine- and human-readable entity 
relationship semantics.

As I said earlier, this is a debate about relations that boils down to 
the following :

1. removing the owl:InverseFunctionalProperty attribute from cert:key -- 
then it no longer includes *explicit* identity reconciliation semantics
2. creating a new relation that has the loser relation semantics that 
you desire.

Having many predicates in an ontology for expressing different kinds of 
relations isn't a bad thing. That's what this whole Semantic Web effort 
is about. You can even make a super property as in:

cert:key rdfs:subPropertyOf cert:superLooserKey .

If you want to "agree to disagree" then you can make:

cert:key owl:equivalentProperty cert:superLooserKey.

It's all about Webby Semantics aka. the Semantic Web :-)

Related:

1. http://bit.ly/Y6TIfs -- using IFP relationship semantics to reconcile 
disparate identities .

-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen