Re: Why contexts? (was: Klyne Contexts: 5.3-5.5 resources, languages and frames)

At 03:56 PM 11/28/00 +0100, Jonas Liljegren wrote:
>Graham Klyne <GK@Dial.pipex.com> writes:
>
> > I believe that resources are too low-level a concept for describing
> > complex systems.  My intuition, which I am planning to demonstrate
> > through examples built over the coming months, is that construction
> > of complex models requires more flexible, higher-level constructs
> > than resources and statements.
>
>I would like to see those examples.

So would I ;-)  I have tried to offer some very simple examples, but the 
nature of the problem is that simple examples don't fully expose the issues 
of complexity.

> > There is one particular issue I have identified when trying to
> > construct complex system descriptions:  it is difficult to construct a
> > description from just nodes and arcs without having an prior knowledge
> > of the complete ontological framework.  I contend that when describing
> > complex systems we do not start out with complete knowledge of the
> > ontological structure that will describe that system.  Development of
> > the statements and the ontology move forward together.
>
>Yes. It's important that RDF data can handle schema evolution.  My
>hope and vision is that RDF will solve the compability problem.  We
>should not have to upgrade the data in orde to use new schemas.
>
>I have taken TBL's concept of conversions [1] as the basis
>for handling schema evolution.  You can inventy your own predicates
>for what you want to say. Someone else can later state that those
>predicates has some relation to other predicates.

[...]

I agree with what you say, and schema matching and conversion is also a 
facet of our research efforts.  But I am talking about a different kind of 
"change".  Schema conversions can work when you have some model and 
complete matching ontology.   I am concerned with the process of getting to 
the point of having a model and corresponding ontology.  I want to make 
statements that are well-formed RDF, that stand independently of the 
detailed supporting ontology.

Given any snapshot of a model, and corresponding (incomplete) ontology, I 
have no doubt that you could add properties (and implied ontology) so that 
contexts-as-statements are not needed.  (This is one reason why I am clear 
that I'm not adding to the logical capabilities of RDF.)  But that's not 
the problem I'm tackling.

I'm not certain I'm following the right approach, but it looks promising 
enough that I think I should do the experiments and circulate some detailed 
examples, then review the results and see if they couldn't be achieved by 
more direct means.

#g


------------
Graham Klyne
(GK@ACM.ORG)

Received on Wednesday, 29 November 2000 16:07:40 UTC