At 03:43 PM 2/8/99 +0800, Simon Cox wrote:
>Karen Coyle wrote:
>>
>> DC has to ... accommodate structurally the metadata
>> that people want to create, and some people will want all of that
>> information in a single piece of metadata because they consider the
>> combination of those pieces of information to be the true representation of
>> the item, not the separation of them.
>
><squeaky_wheel>
>There is no real question that DC can do what is needed *structurally*,
>using DC.Relation, perhaps DC.Source etc. The difficulty is in the
>behaviour of the discovery agents: when confronted with a situation
>where
>significant discovery metadata may be actually distributed between the
>metadata for the item in hand and that for all of its ancestors and
>sources,
>should the index go out and harvest all the values for all the related
>resources
>and use those in responding to queries?
Simon, you're assuming that the burden should be placed (or will be
placed... I'm not sure of your intention here) on the discovery agents. I
think we shouldn't assume that. Instead we should try to find a solution
that is:
1) Relatively easy for humans to input accurately
2) Creates an unambiguous data field where we can know exactly what data
were intended for discovery of which items
In discussing similar dilemmas for the USMARC record, we realized that
there are strong arguments to creating a single record in cases where an
item can have different "facets" that are important for discovery. Separate
records have a host of disadvantages, including the fact that they can
become separated from each other as data is transferred from one system to
another and therefore the desired whole picture of an item can be lost. As
a systems designer, I have to say that I shudder when I know that I will be
receiving "piecemeal" data entries that belong together. And there's
nothing worse than having a record that points to a non-existant "related"
record.
If the aspects of the item are considered equal for discovery, and if they
are not logically viewed as stand-alone, then I see no earthly reason to
place the burden should be on the discovery agent and not record structure.
I prefer efficiency to theoretical neatness.
----------------------------------------------
Karen Coyle [log in to unmask]
University of California Digital Library
http://www.dla.ucop.edu/~kec
----------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|