For a similar perspective to Andy's in another context see the article about
reusing NSDL metadata at
http://metamanagement.comm.nsdl.org/Metadata_Augmentation--DC2004.html
From the article:
--->>>
Abstract:
Digital libraries have, in the main, adopted the traditional library notion
of the metadata "record" as the basic unit of management and exchange.
Although this simplifies the harvest and re-exposure of metadata, it limits
the ability of metadata aggregators to improve the quality of metadata and
to share specifics of those improvements with others. The National Science
Digital Library (NSDL) is exploring options for augmenting harvested
metadata and re-exposing the augmented metadata to downstream users with
detailed information on how it was created and by whom. The key to this
augmentation process involves changing the basic metadata unit from "record"
to "statement."
<<<---
Lorcan Dempsey
http://www.oclc.org/research
http://errol.oclc.org/laf/n90-602202.html
-----Original Message-----
From: Fred Riley [mailto:[log in to unmask]]
Sent: Monday, July 12, 2004 6:40 AM
To: [log in to unmask]
Subject: UK LOM Core: mandatory elements
Interesting post, Andy. I'd tend to agree with your points, although I
hasten to add that I'm a metadata neophyte, coming from a
web/database/e-learning techie background. I'll shortly have to make a start
on creating a learning object repository, which will almost certainly use
the UKLOM Core or parts thereof. I've skimmed through version 0.2 and the
amount of detail required is frightening, although I can see the importance
of most of the elements and wouldn't want to see any dropped. I do worry,
though, about how and who we're going to get to fill in all this data for
each LO, and the number of mandatory elements is also a concern (hell, I
don't even understand some of them, but that's probably because I'm a
programmer not a librarian).
I can also see that sometimes a seemingly obvious mandatory element can't be
filled out. For instance, I remember Niall Sclater in Strathclyde writing
about how the 'title' element of the Question Test Interoperability spec can
be difficult to fill in for a particular question - what do you call yet
another multiple-choice question? I think he wrote that sometimes he had to
fill this with a standard default string. He wasn't moaning, as he figured
the occasional 'clunky' element value was a price worth paying for a decent
schema, but it does go to show that what can seem to be an easy field to
fill in sometimes isn't.
From my limited database design experience, I'm aware that users often run
into problems filling in mandatory fields, but then I'm also aware that when
I made fields optional either users didn't fill them in or just left in
defaults which left considerable gaps in some records.
I'm not drawing any conclusions here as I don't really have the expertise to
do so, I just thought I'd throw in my 2 Euro's worth seeing as you wanted
comments.
Cheers
Fred
Fred Riley
Learning Technologist
Room C57
School of Nursing
University of Nottingham
Queen's Medical Centre
Nottingham
NG7 2HA
Tel: +44 (0)115 92 49924 ext 37180
Email: [log in to unmask]
>>> Andy Powell <[log in to unmask]> 09/07/2004 16:32:40 >>>
My feeling is that UK LOM Core has a major problem with the way it mandates
so many elements.
I think this desire to mandate elements comes from a world view which
assumes that metadata is always passed around in discrete and complete
chunks called records - and that therefore it is always possible to say
whether any particular chunk meets some rules that determine whether it is
valid as a record or not.
More open views, like that of RDF for example, don't take such a fixed view
about whether something is a 'record' or not. They take the view that sets
of metadata properties (elements) can be pulled together from anywhere and
combined in various ways to meet a particular need. Therefore, the notion of
something being mandatory or not simply doesn't arise - if you are missing
an element it may just be because you haven't yet found the bit of metadata
that provides that information yet.
Consider the following scenario...
1) An RDN hub provides the core descriptiove elements about a resource
(title, description, subject keywords, etc.).
2) An LTSN centre provides some e-learning metadata about the same resource
(level, semantic density, learning time, etc.).
3) A third-party service allows end-users to provide annotations about the
same resource.
Each of these services makes their metadata available using the LOM XML
binding. Each follows the cataloguing guidelines in the UK LOM Core for the
elements that they control.
Taken together, the three parts provide a full, UK LOM Core compliant
record. But individually, the metadata exposed by each service is not
compliant because it doesn't contain the full set of mandatory elements.
I think that it would be more helpful to allow these services to claim
compliance with the UK LOM Core, even though they each only contain partial
information. It somehow feels wrong, or at least I don't understand what we
achieve, by saying that the individual services are not complient.
The other argument against making so many elements mandatory is that for all
elements (with the possible exceptions of the identifiers) there will be
some scenarios in which the element has no valid value. In such cases, the
only course of action is to provide the element with a null value? I assume
that a null value is legal(?) but doing this seems, to me, to completely
undermine what is being attempted by trying to mandate particular elements.
Why is an empty value OK but a missing element is not?
In conclusion, I think that the only elements in UK LOM Core that are
mandatory should be
1.1 general.identifier
and
3.1 metaMetadata.identifier
and even those I'd be prepared to be argued out of! Everything else should
be highly desirable, desirable or optional. This would allow applications
to make sensible decisions about which elements they expose or not.
In summary, conformance with UK LOM Core should tell you more about how
particular elements and values have been used than about which set of
elements to expect in a record.
Thoughts?
Andy
--
Distributed Systems, UKOLN, University of Bath, Bath, BA2 7AY, UK
http://www.ukoln.ac.uk/ukoln/staff/a.powell +44 1225 383933
Resource Discovery Network http://www.rdn.ac.uk/
ECDL 2004, Bath, UK - 12-17 Sept 2004 - http://www.ecdl2004.org/
This message has been scanned but we cannot guarantee that it and any
attachments are free from viruses or other damaging content: you are advised
to perform your own checks. Email communications with the University of
Nottingham may be monitored as permitted by UK legislation.
|