Re: Another attempt at 'cascading DRs'

I think this post from Jeremy does clarify the key questions:

Jeremy Carroll wrote:
[..]

> 
> Scenario A) an end user clicks on a link U and the powder agent has to 
> find and apply all relevant trusted powder descriptions, this can 
> involve dereferencing U
> 
> Scenario B) an end user is doing a search and we want to process the 
> uris in a database without dereferencing them.

Scenario B is the one we have in mind for most cases. It addresses the 
important goal of content discovery and is at the heart of all our 
Resource Grouping work.

But... there are some situations - and we refer to T-Online as an 
exemplar of this - where this doesn't work and we therefore need to look 
at something like scenario A. Dereferencing the URI is required whether 
you need to override particular property values or simply find a link to 
a different description, so in either case, we can't, I think, magically 
hang on to the discoverability aspect.

So, to recap a little on the thread, I think we either:

alpha) Devise a means of embellishing a vocabulary to say that if a 
particular property have value X and you subsequently find more data to 
say that it's Y, then use Y. If you already have value Y, this cannot be 
overridden by value X (if something says it's child safe, it can be 
labelled as child unsafe but not the other way round).

beta) Devise a means of defining a description that can be applied to 
any URI within a given host by linking to it.

To my mind, alpha looks fraught with difficulties, beta looks much simpler.

Phil.

Received on Friday, 15 February 2008 09:49:45 UTC