We were asked for a view on what should happen to the IMS metadata spec...
For what its worth my thinking is along the following lines. Apologies,
this is somewhat short and rushed. I don't much care whether this is an
IMS 'problem' or an IEEE LOM 'problem'... but I think our communities
would benefit from a pretty fundamental overhaul of IMS/LOM instance
metadata. I would do this in accordance with the following principles.
- the top level grouping of elements (general, meta-metadata, etc.) within
LOM XML instance metadata serves no real purpose and should be removed...
i.e. the instance metadata should follow a much flatter model
- LOM should adopt the DCMI notion of an 'encoding scheme' and should make
use of existing XML mechanisms (xsi:type) to encode this information (as
per qualified DC in XML proposals)
- LOM should adopt existing XML mechanisms (xml:lang) to encode the
language of metadata values
- all 'identification' within the LOM should be based on the use of URIs
- the meta-metadata section of the LOM should be pulled out into a
separate schema... it is not metadata about the resource being described
- every property within the LOM element set should be given a unique name
(see nesting issue above)
- the underlying 'model' should be enumerated in plain text. What I mean
by this is that there should be some statement about the nature of the
resources being described using LOM, the nature of the descriptions
themselves and the relationships between the resources and the
descriptions. The model should be syntax neutral - i.e. it shouldn't
refer to XML constructs. This would help when attempting to encode the
LOM in new syntaxes (e.g. RDF).
- there should be some cleaner separation between the 'element set' (the
set of properties) and the XML binding associated with it. This should
also separate out issues around the use of AND and OR combinators.
There would be clear benefits to this approach. In particular, it would
allow for much easier combination of LOM and DC into application profiles
and would align the LOM more closely with the Semantic Web (i.e. with
RDF).
Just my 2p :-)
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/
|