Re: [Discuss-DOI] metadata paper #2 version 3: mutability of metadata

[cmorgan@wiley.co.uk]

One aspect of the discussion paper that worries me is that it implies that
metadata is to be collected on registration of the DOI and that key
(required) metadata are immutable. (Can't be altered.)

>Does it really say this? I thought that the reference to authority issues
(section 5.3.1) meant that there could be later versions.

It became clear at the NISO linking meeting last week that for the purposes
of linking, publishers want a work identifier for scholarly articles that
is unique and unitary. For every scholarly article there ought to be one
and only one identifier. (DOI now provides us with the "one", but not the
"only one", function.) However, if key metadata are immutable, then a new
DOI would have to be issued if key metadata had to be changed. Right now,
the discussion paper makes a big point of requiring the principal creators
to be exhaustively enumerated in a key field.

>I think we need to make a distinction between key metadata that you use to
describe the resource itself and key metadata that you need to enable
linking. That is to say, there are various metadata elements that everyone
can agree would be key descriptors of a resource, e.g. (using DC Simple
elements rather than DOI/INDECS ones) DC.Creator, DC.Title, DC.Identifier.
But these elements may not exist in citations (e.g. the ISSN is clearly key
metadata that will be attached to a resource, but no citation ever quotes
it) or they may exist in citations but not be necessary for linking (e.g.
the article title is probably redundant for linking information, and you
are unlikely to need more than the first author). Citation metadata clearly
has to have some association with resource metadata, but it is not the same
metadata set, and we shouldn't confuse "bare-bones" resource metadata with
"bare-bones" citation metadata. (Maybe another example will help to
clarify: we may all agree that DC.Publisher is a key piece of resource
metadata, but it is completely irrelevant for linking purposes, and it is
never quoted in journal citations. So a bare-bones resource metadata set
would include Publisher, but a bare-bones citation metadata set wouldn't.)

>Cliff Morgan

>Publishing Technologies Director

>John Wiley & Sons Ltd

>Chichester, UK







------------------------------------------------------
Discuss-DOI maillist  -  Discuss-DOI@doi.org
http://www.doi.org/mailman/listinfo/discuss-doi