During the Metadata Interest Group session this year’s ALA Annual Conference Erik Mitchell, Associate University Librarian at UC Berkeley, gave a presentation titled, “Trends in Linked Data Adoption,” based on his recent research.
Erik began his presentation by describing the model he used for evaluating metadata systems (citing Elings and Waibel’s terminology); this model provided a framework for comparing different approaches by three of the largest linked data initiatives in libraries to date (DPLA, BIBFRAME and Europeana). The Digital Public Library of America (DPLA) approach to metadata includes harvesting, aggregating, republishing (similar to OAI), with some normalization for faceted access. BIBFRAME is still in the data modeling phase, and (because it is seen as a replacement for MARC) BIBFRAME is largely focused on serving as an exchange standard. Europeana uses a common schema, index, and dissemination system to bring together European libraries, museums, and other collections.
Erik elaborated on other methods (community discourse analysis, API use, and metadata visualization) he has used to gain insights into the three library linked data initiatives. In his early discourse analysis, Erik looked at word counts for different documents generated at the NISO Bibliographic Roadmap meeting and was able to break discussion up into themes. He also pointed out that relevant social media spaces and listserv content would be a good candidate for similar analysis. Erik explained how APIs for DPLA, BIBFRAME and Europeana are showing an increased overlap in serialization and exchange standards. The presentation also included a quick mention of a number of data visualization tools, namely Gephi and W3C’s RDF validator, that can be used to find implicit and explicit connections between different linked data. During the Q and A Jenn Riley offered Turtle as another option for visualizing linked data.
There was lively discussion during the session’s Q and A period about whether libraries should be migrating or mapping existing records to linked data formats and the pros and cons of reusing or minting new vocabularies. There seemed to be a preference amongst many in the room for an eventual migration of our data to linked data formats and for the reuse of existing vocabularies, but libraries are still trying to decide which data model/s, vocabularies, and format/s to develop and adopt.
Erik has a related publication due out in Library Technology Reports; slides for the MIG presentation can be found at: http://connect.ala.org/files/ALCTS_MIG_ALA2013_Mitchell_20130630.pdf.