Re: SIMILE PI phone conference, 26-Sep-03 1200 EDT/1700 BST

Hi Mark and MacKenzie,

I should say my reason for pointing to the article was that it articulates
some of the issues that were identified as part of the ARKive-ERA
(educational repurposing of assets) project about using  'educational'
metadata (e.g. IEEE LOM) to mark-up 'raw' multimedia data e.g. images, short
video clips that do not themselves have educational goals intended.

The major point was exactly that stated by MacKenzie, that in any context
where you have to choose from a finite set of metadata elements and
discriptive vocabularies (i.e. always, given current technologies), that
choice is [has to be] contextual. It is dependent on assumptions about how
the object is to be used and by whom - the dilemma being that ideally data
providers probably don't want to have to second guess how their, hopefully
very diverse, users will want to use it.  As I say in the article, I'm not
aware of any single or simple solution to that dilemma, as yet. But I think
domain ontology mapping is one potentially valuable tool in a larger
tool-set.

The elements I pulled out of the LOM standard were specifically those
related to 'pedagogic characteristics' ('use metadata' as defined by the
Gilliland-Swetland reference) because those were the focus of the article.
But those elements are only one (relatively small) sub-set of the LOM
elements, the majority of the others are broadly speaking, more generic.

Almost by definition (?) specialist metadata standards, (i.e. other than
those like DC that are specifically designed to be generic), will have such
'use' elements that are specific to their application/subject area -
including VRA e.g. Date.Restoration or style/period. As MacKenzie says,
perhaps it is just that some are more explict than others. How to map those
(or not) is a really interesting and important problem esp. in the context
of SIMILE, where in the longer term there will be very many element sets and
vocabularies.

I think the current example for the demo is a good and realistic one -
educational users will want to use IMS/IEEE LOM metadata to mark-up 'raw'
images - even though it is not ideal for that purpose, because that's what
the Virtual Learning Environments and their other tools, are likely to use.
However a specialist image library (of interest to academics) would probably
use something more like VRA. Two way interoperation is likely to be a
necessity if those two communities are to get collective and mutual value
out of their respective collections.

I think (and its late, so I might not be thinking clearly) that if you go
from one to the other (IMS to VRA and/or vice versa) you will *have* to
loose some information. So perhaps the trick we want to learn is how to
minimise that loss and still end up with valid, meaningful and interoperable
metadata at the end of it. In that context the discussions of intermediate
schemas seems like an interesting area...

Paul

----- Original Message -----
From: "MacKenzie Smith" <kenzie@MIT.EDU>
To: "SIMILE public list" <www-rdf-dspace@w3.org>
Sent: Monday, September 29, 2003 10:29 PM
Subject: RE: SIMILE PI phone conference, 26-Sep-03 1200 EDT/1700 BST



Hi Mark,

I haven't read Paul's article yet (I'll get to it soon) but I'd like to
respond to
this thread right away.

>As your paper correctly notes, there is a huge difference between metadata
>that describes how we might use a resource and metadata that describes a
>resource.

Actually, in the world of metadata producers (aka catalogers) it's pretty
well
understood that there's absolutely no such thing as objective descriptive
metadata. *All* metadata is informed by context, world-view, usage scenario,
educational background, and other attributes of the person supplying the
metadata for a particular purpose.

>The problem we face here is IMS is a "use" metadata schema for an
>educational context, whereas VRA is a descriptive metadata schema.

I disagree. IMS is descriptive metadata for an educational or pedagogical
context. VRA is descriptive metadata for a more general, research-oriented
context. So some elements are shared and others are unique to each schema.
I don't think the distinction you're trying to make between "use" and
"description" is useful -- description is provided as a short cut to help an
intended audience, whether that audience is identified explicitly or not.
Maybe usage information is just explicit in IMS while it's implied in VRA.

Which isn't to say that we won't toss out some IMS elements which are
*so* specific to the educational/pedagogical context that they have no
value in any other domain. It might even turn out that the useful subset
of the IMS schema is approximately equivalent to Dublin Core. But I
don't believe that VRA and IMS are that fundamentally different:
description and context and use are really inseparable...

Best,


MacKenzie/



MacKenzie Smith
Associate Director for Technology
MIT Libraries
Building 14S-208
77 Massachusetts Avenue
Cambridge, MA  02139
(617)253-8184
kenzie@mit.edu

Received on Monday, 29 September 2003 21:09:07 UTC