Quoting "Brenndorfer, Thomas" <[email protected]>:
I want to thank Thomas for this detail and yet clear description. I
think we need to capture explanations like this somewhere, since email
is such a transitory communication device.
I want to add something to his section on linking elements:
Then there are conventions for entering the value for these elements:
- identifiers
- authorized access points (not used for manifestations and items)
- structured descriptions
- unstructured descriptions
Linking can be "actual" or "virtual." By that I mean that there can be
actual links between machine-readable entries, or there can be enough
information that a person can make the mental leap to understand the
link, but a machine cannot. For machine linking we really need to have
identifiers for the originator (e.g. the Work being described) and the
target (the work being linked to). Controlled strings of text can be
identifiers, but that only works within a closed system, like a single
database. If you want to have your data link accurately in an
environment with data from a mix of sources, as in a consortial
database, then you have to know for sure that your string uniquely
identifies the element in that context, e.g. that it exactly matches
the authority entry being used.
One of the difficulties in implementing this in MARC is that, as
Thomas showed, there isn't always a direct connection between what is
in a MARC linking field and the thing it will link to. While it seems
logical that a 7XX link to an authority 100 with $t (etc.), it cannot
link directly to a 1XX/240 combination. The intermediate steps needed
to implement such a link mean that the linking function requires a
bunch of data normalization, and thus has to happen within a certain
application environment. There are also differences between the
subfielding in 7xx's and other heading fields that can interfere with
matching.
I believe it would take some serious adjustment of MARC to make actual
linking work, if it is possible at all (it would take some study to
come to a clear conclusion). Yet it is this ability to create links
that is new in RDA. Unfortunately, using MARC we can't try it out to
see if linking gives us some advantages that might make RDA worth while.
--
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
