- From: Karen Coyle <kcoyle@kcoyle.net>
- Date: Tue, 14 Sep 2010 11:50:58 -0700
- To: public-lld <public-lld@w3.org>
Quoting "Svensson, Lars" <l.svensson@d-nb.de>: > > I agree in general, too. Would a possible solution for the expression > level be to create an ad hoc-expression for each manifestation, just in > order to have one? That way you cannot build a good user interface _at > once_ but eventually you could (intellectually) merge the "records" > describing the same expression. That would definitely be a human task > (crowdsourcing?). I think my situation brings up the dilemma: what is the nature of that expression "level"? It seems to me that we have (at least) two options. I need to draw this out in some detail, but in general it seems that these two sketch out as: 1. separate identification each of WEMI for a given bibliographic resource; each will have a URI, and properties will be associated only with one of the WEMI URIs (this is the RDA viewpoint) Work123: Title of Work123 Author of Work123 Expression456: Expression of Work123 Language of Expression456 Manifestation789: Manifestation of Expression456 Title of Manifestation789 place of publication of Manifestation789 Publication date of Manifestation789 etc. 2. identification of a bibliographic resource as a whole; a URI for the resource; with interpretation of the "FRBR-ness" of properties by applications, probably based on an application profile. This would allow some views of the data that vary from the strict FRBR view.(Note that in RDA, the properties are all associated with a single FRBR entity, so there is no ambiguity between them in terms of FRBR association in an RDA environment.) BibResourceABC: Work title of BibResourceABC Author of BibResourceABC Language of BibResourceABC Manifestation title of BibResourceABC Place of publication of BibResourceABC Publication date of BibResourceABC Works, Expressions and Manifestations could be identified *somewhere* (e.g. Open Library, Freebase), and probably in more than one place. If, however, there were no identified entity, this model would still be functional. Some entities also have commonly known identifiers (and possibly more than one) that could help bring together data for the same entity but generated in a different environment. (This is actually how RDA defines identifiers for WEMI. BibResourceABC: BibResourceABC represents FreebaseWork123 BibResourceABC represents OLWorkXYZ BibResourceABC represents Expression456 (identified with ISTC ID) BibResourceABC represents Manifestation789 (identified with ISBN, LCCN...) Work title of BibResourceABC Author of BibResourceABC Language of BibResourceABC Manifestation title of BibResourceABC Place of publication of BibResourceABC Publication date of BibResourceABC Identifiers would be very useful for making statements like "Expression456 is a translation of FreebaseWork123." But that does not mean, to me, that the Work or Expression are "records" in the sense that they appear to be in FRBR; that is, that the properties are related structurally to that identifier. (And note that in FRBR there is no identifier for the whole bibliographic description, something I find odd, and which creates difficulties if you cannot derive a full FRBR entity set from your data, such as my dilemma of having a Manifestation and a Work, but no Expression identified to bring them together.) This #2 is a realistic version of the data that we have today, pre-FRBR. Remember, however, that FRBR is a new way to conceptualize the bibliographic data that we already create -- it is structurally new, but most properties are not new. Today's data is not organized in terms of FRBR entities, but most (all?) of the data in our records has a place in FRBR. I realize that this looks like a re-writing of FRBR, but in fact FRBR, as Gordon reminds us, is a CONCEPTUAL model not a data model. To arrive at a data model we need to think about how we want to use the data and what services we wish to provide. I happen to think that in many if not most cases, the bibliographic resource will be treated as a whole for retrieval (searching), and often will be displayed as a whole. I appreciate the efficiencies that the one-to-many relationships may make for data creation, but the information discovery function may not be best served by that same structure. Different applications can use the many bibliographic properties differently. kc -- Karen Coyle kcoyle@kcoyle.net http://kcoyle.net ph: 1-510-540-7596 m: 1-510-435-8234 skype: kcoylenet
Received on Tuesday, 14 September 2010 18:51:33 UTC