RE: "Metadata about metadata" paper published in Future Internet

Peter,

Thanks for comments/questions.

I'll cross-post a reply here and to my blog on the portaldoors.org web site before I leave on Friday to attend HealthGrid 2010 in Paris.

CT

-----Original Message-----
From: public-semweb-lifesci-request@w3.org [mailto:public-semweb-lifesci-request@w3.org] On Behalf Of Peter Ansell
Sent: Monday, June 21, 2010 4:17 PM
To: ctaswell@telegenetics.net
Cc: public-semweb-lifesci@w3.org
Subject: Re: "Metadata about metadata" paper published in Future Internet

On 22 June 2010 02:30, Carl Taswell <ctaswell@telegenetics.net> wrote:
>
> The paper entitled “A Distributed Infrastructure for Metadata about Metadata: The HDMM Architectural Style and PORTAL-DOORS System” has been published in 2010 Future Internet 2(2):156-189.
>
>
>
> Future Internet is an Open Access journal, and the paper is available for download at
>
>
>
> http://www.mdpi.com/1999-5903/2/2/156/
>
>
>
> Questions, comments, suggestions and criticisms always appreciated….
>

Sounds very interesting. I don't think I have it fully understood at
this point, but I am just wondering about a few points right now.

You say on page 171 that "All labels, whether canonical or alias, for
all resources must always be globally unique throughout the system.
Thus the original design requirement for uniqueness of labels has not
been violated by this revision."

I am wondering how a heirarchical distributed system manages to ensure
this requirement is met. If I say that the canonical identifier for
something is <myuri> and someone else says it is <theiruri> how does
the system negotiate this? Do the registries have to be connected
before the second URI is assigned?

Is there a long startup time for adding new registries and new
resources where everyone has to relearn everything about everyone else
to make sure things are still consistent? You say in the scenario that
different scientists should be free to independently say things about
the same thing. Does that thing have to be given a label by a single
and final authority initially or could the system gradually learn what
the preferred resource label is?

Also, what are the approximate number of resources that you were
managing in your trial, and how many aliases were used for each
resource on average?

Cheers,

Peter

Received on Tuesday, 22 June 2010 21:54:39 UTC