Feb 25, 2014
Question from Stefanie of whether we will ever be able to move to non-flat metadata. Example of dcterms.publisher and need to use URI, as specified in domain and range. If only providing a string, will need to use dc.publisher rather than dcterms.publisher. Bram suggests could store a URI but would need functionality of human lookup name for that URI, for display. These values would not necessarily reside in the metadata but in some other index. Maureen suggests need to supercede and maybe replace the old mappings-- more current mapping would indicate that both DCTERMS and DC ship with DSpace. Do need to address the mapping document.
Bram asks about QDC in the DC community. Stefanie reports that most projects use DCTERMS alongside DC. Did some work on application profile for RDF representation, published this year, using DC, DCTERMS, and terms from other vocabularies, like VIVO, to describe objects. Use of MARC relator codes (http://id.loc.gov/vocabulary/relators.html) and URIs for qualified contributors. References user guide: http://wiki.dublincore.org/index.php/User_Guide Need a mix&match approach, ability to pull in terms beyond Dublin Core. Bibframe, replacement for MARC, working on vocabularies for bibliographic data. Focus on work & instance as a version/simplification of FRBR. In RDF: qualifier = subproperty. MARC relator code expressed as subproperty of creator or contributor.
Discussion of whether RDF needs to be integrated into DSpace internal format or not. Possibility, instead, of capability to crosswalk RDF out of internal format. Stefanie indicates that this is a matter of preference-- some prefer that the internal format of the database be distinct from the export format (makes it more flexible to transform to DC, DC RDF, MODS, etc.). Not absolutely fixed on the format-- more flexible and allows for transformation. But have to make sure that what is in the data is what you really need to generate RDF. Need to be able to store URIs and MARC relator codes in internal format. Others say it is easier to keep data in the format you intend for distribution. Points us to German National Library and data transformation efforts (https://wiki.dnb.de/pages/viewpage.action?pageId=68061169). Recommendation is "rather flat" but seen as "first step on our way to good linked data having in mind the Europeana Data Model... and BIBFRAME."
Stefanie will look at changes her colleagues have made to the original DSpace registry.
For Dublin Core, possibility of discussing this proposal at the Libraries community meeting.