Two Paths to Interoperable Metadata

http://www.oclc.org/research/publications/
archive/2003/godby-dc2003.pdf

It used to seem that the problem we had was not having enough metadata. Now it is becoming clearer that not only do we have a lot of existing metadata, we have a lot of competing metadata standards and schemas – different ways of describing what are either the same or similar objects.

This paper, presented at the recent 2003 Dublin Core conference, proposes one (in fact two) model to translate between this proliferation of schemas in order to present users with a unified search interface to diverse collections tagged with different metadata schema.

The approach advocates the need to decouple syntactic transforms from semantic mappings through the creation of a web services-based “repository of crosswalks” – essentially a database of equivalencies between the schemas elements created by human experts. Translations between schemas would be facilitated through the notion of an ‘interoperable core,’ a set of core elements that had been mapped to the equivalent elements in the various schema.

It would seem like such an approach is stepping into the territory claimed by RDF, but the authors counter this approach with in one line: “We believe that the interoperable core could be implemented in RDF, but our experience has led us to conclude that RDF representations introduce unnecessary processing bottlenecks.”

The paper ends by exploring the suitability of any existing schema as the ‘interoperable core’ and concludes that there may be multiple cores, depending on whether one is trying to translate within one “community of practice’s” schemas (e.g. the library world, with its MARC, EAD or ONIX) or between different community’s schemas (e.g. between the community of learning object developer’s LOM and the librarians MARC). – SWL

– via [ResourceShelf]