Pete--

I'd go with option b):

>(b) the capacity to record a relation between the current collection and
>an associated collection, where the nature of the association is
>unspecified (or something loose like the EAD relatedmaterial definition
>- "may be of use to a reader/user")?

In our project we're using this to let National Leadership Grant projects
make connections between the collection being described and a complementary
or otherwise associated collection -- really a sort of catch all
category.  So it would make sense to just use dc:relation.

Sarah

At 02:36 PM 8/29/2003 +0100, you wrote:
>(These various threads about relation attributes have reminded me that I
>meant to highlight this point and ask for comments.)
>
>The RSLP CD schema property hasAssociation is defined as
>
> > The identifier or name of a second collection that is associated by
>provenance with the current collection.
>
>and that same definition is used in our current proposal.
>
>Note that this is not just any association but specifically association
>by provenance in the sense that term is used by archivists:
>
>http://www.ica.org/biblio/cds/isad_g_2e.pdf
>
> > The relationship between records and the organizations or individuals
>that created,
> > accumulated and/or maintained and used them in the conduct of personal
>or corporate
> > activity.
>
>I think this property was probably included in the RSLP CD schema in
>order to provide a mapping to one or both of
>
>- the Associated Material (3.5.4) element in ISAD(G) 1993 version
>("to indicate the existence in other repositories of material associated
>by provenance to the unit of description.")
>- the separatedmaterial element in EAD 1.0
>("information about materials that are associated by provenance to the
>described materials but that have been physically separated or
>removed.")
>
>Note: EAD 1.0 has another element relatedmaterial ("information about
>materials that are not physically or logically included in the material
>described in the finding aid but that may be of use to a reader because
>of an association to the described materials. Materials designated by
>this element are not related to the described material by provenance,
>accumulation, or use.")
>
>In practice, I strongly suspect that many implementers of the RSLP CD
>schema made a somewhat loose interpretation of the hasAssociation
>property, and used it to record relations between collections based on
>various criteria that might be of interest to a user, and not only
>association by provenance.
>
>In the 1999/2000 revision of ISAD(G), the distinction between what had
>been called "Related units of description (3.5.3)" and "Associated
>material (3.5.4)" was removed and there is a single element:
>
>- Related units of description (3.5.3), covering, "information about
>units of description in the same repository or elsewhere that are
>related by provenance or other association(s)."
>
>i.e. the relationship is no longer limited to association by provenance.
>
>EAD 2002 does continue to provide two distinct elements as above.
>
>So (with apologies for the long-windedness, but I thought the background
>needed explanation), do we need either or both of:
>
>(a) the capacity to record a relation between the current collection and
>an associated collection, where the association is specifically by
>provenance?
>(b) the capacity to record a relation between the current collection and
>an associated collection, where the nature of the association is
>unspecified (or something loose like the EAD relatedmaterial definition
>- "may be of use to a reader/user")?
>
>I'm inclined to say we want (b) but I'm less sure we really need (a).
>What do others think please?
>
>And if we do require (b) and not (a) (i.e. our concept of "association"
>is loose), do we need to define a specific hasAssociation property, or
>should we just fall back on using dc:relation (with a label ("Related
>collection" or "Associated collection") and a use definition specific to
>this profile, as for some of the other DC properties)?
>
>Pete