Page 183 - Building Digital Libraries
P. 183

CHAPTER 7


                                                   Within organizations, metadata practices will change over time. Shifting
                                                   data models and formats have exacerbated this problem, as institutions
                                                   reevaluate data scope or migrate to new digital library systems. These types
                                                   of changes, while possibly advantageous within a local system, represent
                                                   barriers preventing an organization from participating in larger interoper-
                                                   ability efforts.

                                                   Schema Granularity
                                                   Very rarely does crosswalking occur between two metadata schemas that
                                                   share the same level of granularity. In part, this is due to the nature of
                                                   descriptive formats—they tend to be created to meet a specific need and
                                                   capture specific types of descriptive information. As a result, the crosswalk-
                                                   ing of metadata becomes a process of (1) deciphering how metadata ele-
                                                   ments are related, and (2) determining whether the crosswalk will result in
                                                   a loss of data granularity. In the best-case scenario, metadata crosswalkers
                                                   are looking for metadata elements that have one-to-one matches; that is, a
                                                   title in one schema that matches directly with the title element in another
                                                   schema so long as the granularity within the title elements is equal. In this
                                                   case, metadata can be directly mapped from one element to another. This
                                                   would be a lossless crosswalk—something really only achieved when mov-
                                                   ing data between MARC21XML and MARC. More likely, metadata cross-
                                                   walkers will need to deal with one-to-many and many-to-many matches.
                                                   This is where one metadata element in one schema matches the definition
                                                   of multiple metadata elements in a different schema; for example, moving
                                                   data between Dublin Core and MARC21. As noted in past chapters, Dublin
                                                   Core is a relatively low granular format utilizing only fifteen core metadata
                                                   elements. When crosswalking between Dublin Core and MARC, we see
                                                   a number of instances where a single Dublin Core field maps to multiple
                                                   MARC fields (see figure 7.4).
                                                      If we take a closer look at the Dublin Core creator field, we see very
                                                   clearly how granularity can affect metadata crosswalking. The Dublin Core
                                                   creator element, by definition, stores information relating to the publishers
                                                   of a work. This means that any individuals or organizations responsible for
                                                   the creation of the document should be tagged as a creator within the Dublin
                                                   Core schema. Within MARC21, entities are tagged according to the type of
                                                   entity (personal, corporate, etc.) and their relation to the work, that is, as
                                                   the main entry or contributor to a work. As such, the Dublin Core creator
                                                   element could theoretically map into one of seven different metadata ele-
                                                   ments within MARC21. What’s more, since Dublin Core doesn’t capture the


                                                    DUBLIN CORE                    MARC21
                                                    Creator        100, 110, 111, 700, 710, 711, 720

                                                   FIGURE 7.4
                                                   Dublin Core => MARC21 Author Crosswalk



            168
   178   179   180   181   182   183   184   185   186   187   188