Roland,
This is what I suspected and it has a great impact
on real use of external authority data/controlled
vocabularies
Decision, obviously, need to be on the assumption that
external authority data or controlled vocabulary is
permanent namespace.
Or one can make assumption that this is never the case.
> <rdf:Description rdf:about="urn:abc17">
> <vCard:FN>Aida</vCard:FN>
> </rdf:Description>
This is kind of redundancy one would like
to avoid. Especially if eventually "urn:abc17" keep
changing some properties/attributes/relationships.
> The punch line of dcq-rdf/xml is a receiving application should
> be able to do a smart dumbdown not worse than you would do
> by yourself
> for instance in a search interface, which allows a user to search
> authors by names.
>
However, names are not as big issue as
KOS (knowledge organization systems) e.g. those
listed under encoding schemes LCC, DC, UDC.
That is - if one can't rely on having this data externaly.
This means that if KOS is thesarus, for instance, one would need not only
to
declare "encoding scheme" but also have the following accommodated in dc:subject:
Term CAR from 'Mythesaurus'
CAR
UF: automobil
BT: motor vehicle
NT: sport car
RT:lorry
SN: car in mythesaurus comprises blah blah blah
Or...
Term CAR from 'My_company_thesarus'
CAR
UF: automobil
BT: company fleet
NT: personally allocated cars
RT: rented car
SN: car in our company thesarus comprises blah blah blah
Or at least broader term (BT) and narrower terms (NTs), would be needed.
Analogy would be that if one would use DDC, UDC, LCC classification, one
would also need not only to declare encoding scheme and have classification
numbers as value, but description of that number or indexing terms belonging
to the number. And also broader and narrower category if one has any intention
to use this for browsing.
This is highly relevant if one would want to use encoding schemes in DC
subject in information retrieval (search and browse).
The way it is suggested now: to slot in classification number or any other
KOS value, and declare encoding scheme does not make sense - from the point
of view of the function this is meant to perform.
This solution in dc:subject is equivalent as if someone would do the following
in dc:creator :
<dc:creator scheme="system_written_on_paper_accessable_on_demand">17LC33</dc:creator>
I allow that it is necessary to declare 'encoding schemes', but I still
think something is missing here...not in terms of what DCMES, being a standard,
itself needs to solve, but in terms what it ought to allow and anticipate.
One needs to declare rules/KOS and to have real encoding schemes that are
instantiation of these rules that may be dealt with as permanent namespaces.
If there is an intention to share KOS tools it may not be enough to say
scheme="DDC" but may be needed to say that this is scheme="DDC_XY", and
have the possibility to deal with this as a namespace with each indexing
term having URN.
Aida
|