Re: BP 7, provide unique identifiers

Following the chain back to the use cases, it seems the real requirement is for predictable and persistent identifiers. I would support a BP about choosing predictable and persistent URIs.
--
Annette Greiner
NERSC Data and Analytics Services
Lawrence Berkeley National Laboratory
510-495-2935

On Jan 22, 2015, at 2:12 PM, Annette Greiner <amgreiner@lbl.gov> wrote:

> Looking at BP 7, I don’t understand the idea. I’m wondering what we are defining as a “data resource”. From the text, it seems we are using it in the sense of whatever is linked to. That seems fair, because datasets can be available as a single file or several files or by separate API requests for very specific queries, etc.  So, if a data resource is a data element that is linked to, then it is represented by a URL. The discussion of URNs is clearly out of scope for us, as they are not on the web. That leaves URLs, but then the best practice comes down to ensuring a unique URL for whatever has a URL. I think it would be impossible to not do that an still put something on the web. In the context of data on the web, I don’t think much of this best practice is in scope, and the bit that is in scope is tautological.
> -Annette
> 
> --
> Annette Greiner
> NERSC Data and Analytics Services
> Lawrence Berkeley National Laboratory
> 510-495-2935
> 

Received on Thursday, 22 January 2015 22:30:33 UTC