>>I thought, we do not want to have a meta-data about figures in MARC ... >>rather standalone BibDoc remembering everything, but if we wanted to create >>one record per figure, indeed this would be solved. >>(Again brings the question if we want to multiply our record space by 3)
>...that's MARC multilevel description! (in MARC21 dialect: >http://www.loc.gov/marc/bibliographic/bd76x78x.html) If You mean referencing MARC records of figures from the main article document, I think there were some more pragmatic issues with this - the Invenio ability of dealing with ~10000000 records (The number of records would be around this) - separation of records describing publications and other objects (In Inspire we currently have records only for publications) - What would purely figure-MARC store ? All the important information would be stored in tie BibObject (earlier BibDoc). the advantage of having records for figures would be ease of reutilising current search metaphores ... though only for textual data. I will try to answer the longer e-mail soon. (After understanding ... I am not a librarian and it is difficult to understand all related standards and aspects of them at once) I would just like to note that in what I understand, a descriptive meta-data of figures, which would change just as often as MARC records, would be encoded in METS. I will have a look at eamples you mentioned. I think, the biggest question at the moment is: Do we need to change the proposed data-model (Bibobject, BibFiles etc...) to allow importation of METS descriptions. I can see several issues with this, for example we do not allow arbitrarily nested divisions. just document->files (with the dimension of versions). How should we deal with custom meta-data ? you can enclose arbitrary RDF inside METS, while we were so far talking about a dictionary of depth 2 structure. Piotr
