On Mon, 2007-03-26 at 15:11 +0200, Tor Arne Dahl wrote:
> Thanks for all the answers to my questions about profiles and context set.
> On 22.03.2007 15:39, Mike Taylor wrote:
> > Tor Arne Dahl writes:
> > > 1. The profile is based on the NorZIG Z39.50 profile, and we design the
> > > SRU profile to support for the same kind of queries as the Z39.50
> > > profile. Many of the necessary indexes are defined in the Bath, CQL,
> > > Dublin Core and Record Metadata context sets, but we had to create a
> > > NorZIG context set to define all required indexes (series title and
> > > corporate and conference name, for example). Since our first draft, the
> > > CQL Bibliographic Searching Proposal has been published on the SRU
> > > website. Many of the indexes from the NorZIG context set are defined in
> > > the bib context set described in this document, so we would rather use
> > > them than create our own.
> Some of the indexes seem to be included in the proposed bib context set:
> norzig.creatorConference -> bib.nameConference
> norzig.creatorCorporate -> bib.nameCorporate
> norzig.titleSeries -> bib.titleSeries
Yep. with bib.role=creator
> norzig.docid and norzig.nationalBiographyNumber can probably be replaced
> by dc.identifier with the bib.identifierAuthority relation modifier.
> Which values could be used for the modifier in these cases?
norzig.docid would be rec.identifier -- the identifier for the record,
not the identifier for the object being described by the record. Yes?
> norzig.dewey, norzig.remoteSystemClassificationNumber and norzig.udc can
> probably be replaced by bib.classification and some kind of
> bib.classAuthority relation modifier. Are there any suggestions about
> how this can be done?
bib.classification =/bib.classAuthority=dewey "term"
bib.classification =/bib.classAuthority=local "term"
bib.classification =/bib.classAuthority=udc "term"
> The last three indexes specify normalized forms of personal names (i.e.
> last name, first name(s)). This isn't mentioned in the CQL bibliographic
> searching proposal. Do you think it's a bad idea to create these indexes?
The indexes should be normalised anyway?
> I'm also wondering about the dc.creator index in CQL bibliographic
> searching. With the different name indexes and relation modifiers
> defined in the bib context set, does it mean that dc.creator never
> should be used?
dc.creator is useful as a very general index as the semantics of the
element from the dublin core metadata specs are quite broad. If you
want to be more specific, then the bib indexes are useful.