Re: Graphs vs. Authorities vs. Warrants vs. Authentication vs. Certification etc.

On Mar 19, 2004, at 10:10, ext Jeremy Carroll wrote:

>
>
> I basically agree that the Warrant makes the modelling clearer, at the 
> expense
> of making it more complex and detailed. As much as possible we should 
> avoid
> going into details of any scheme (PGP or X.509),

Agreed. I put that in there simply for the example. The less said about
specific methodologies the better -- so long as the message that the 
most
important/widely used ones will be fully supported gets through.

> (although it would be
> helpful to have a few example properties etc. and some idea of how it 
> is done
> - we can even refer to some worked example on the web or something - 
> but it
> is at the wrong level of detail - we have to be convincing that that 
> level of
> detail does not present problems). Agree wholeheartedly with 
> extensibility as
> goal (although that makes MT grounding hard/impossible)

The MT would only, I think, concern itself with the swp: terms and 
criteria
such as "a warrant is valid if the specified authority is authenticated
and the signature checks out" -- and the additional properties 
describing a
warrant simply allow that to happen. In some cases, no further 
properties
are needed. In other cases, proprietary properties are needed. But it 
falls
below the "opacity level" insofar as the MT we define is concerned.

Yes?

>
> Two issues about the cardinality:
> 1) your OWL is incorrect - unimportant I can fix that later.

Oops. No surprise.

Please point out what's wrong so I can avoid similar mistakes in the 
future.

>
> 2) there is also an inverse cardinality constraint on swp:warrant, 
> i.e. a
> warrant has exactly one graph,

Right.

> hence I suggest inverting that property making
> it
>
> swp:graphOfWarrant     @@ change name
>     a rdf:Property ;
>     rdfs:comment "The subject is a warrant by which the object graph 
> can
> be authenticated and
>                   its assertional status and asserting or originating
> authority determined." ;
>     rdfs:range rdfg:Graph ;
>     rdfs:domain swp:Warrant ;
>     owl:cardinality "1"^^xsd:int . @@ sic

Looks OK to me.

Though perhaps just swp:graph is enough, since the domain/range make
the rest clear.

(I tend to prefer having predicates that closely match the range. I've
found that users grok an ontology much better when things line up
mnemmonically in that fashion)

> More on the use of datatypes later ... I am unconfortable really.

OK. I look forward to understanding what concerns you.

Patrick


>
> Jeremy
>
>
>
>
>

--

Patrick Stickler
Nokia, Finland
patrick.stickler@nokia.com

Received on Friday, 19 March 2004 03:26:54 UTC