Re: POWDER: thoughts

That looks like a pretty good summary to me.

Point 1 could be dropped if it is in direct conflict with the others but 
it has been something we've striven for. I would also expand a little on 
8 to say that optimisation within a specific environment is important. 
For example, once you know you have POWDER and you know you trust the 
data, you should be able to process that data efficiently 9this is 
particularly important in the mobile space of course, a major input to 
the WG).

We have a WG call shortly during which I'll refer to your various posts 
and try and come up with a sensible set of questions and thoughts to 
bring to Bristol tomorrow.

Phil.

Jeremy Carroll wrote:
> 
> 
> Hmmmm, trying to think constructively.
> 
> A loose stab at the requirements for these aspects of the design
> 
> 
> 1) The POWDER file should be fairly easy to write, by hand if necessary, 
> and not require a specialised tool.
> 
> 2) The POWDER file should have a formal semantics.
> 
> 3) The POWDER file should have an operational semantics.
> 
> 4) It should not require too much specialist skill to implement the 
> POWDER operational semantics.
> 
> 5) The divergence between the formal semantics and the operational 
> semantics should be small.
> 
> 6) It should be possible to access (most of) the formal semantics of the 
> POWDER file using off-the-shelf semantic web tools.
> 
> 7) Within the off-the-shelf semantic web tools paradigm, POWDER has lots 
> and lots of extensibility (like RDF).
> 
> 8) However, not all that extensibility needs to be apparent in the 
> operational semantics that specific POWDER tools might be using.
> 
> ====
> 
> So, as an example, while in RDF and OWL we can declare a subClass of 
> wdr:ResourceSet, a POWDER specific tool might just use an XML parser, 
> with no Semantic Web capabilities, and hence ignore such a subClass 
> while treating wdr:ResourceSet in a special way.
> 
> Jeremy
>

Received on Monday, 17 December 2007 15:57:33 UTC