NOTE: apologies for the belated posting!
The ALA TechSource webinar, Libraries and Linked Data: Looking to the Future, was presented by Karen Coyle on July 19, 2012. Jennifer live blogged this event.
Coyle begins by giving an overview of linked data principles but focuses on the first rule in particular: use URIs for names of things. Coyle likens linked data triple statements to “things and strings,” or things and relationships between things. She explains that creating identifiers is the first step in taking library data from a flat file (currently serialized in MARC) to a relational/object-oriented database model (a model that RDA aspires to support as a content standard) and eventually to linked data.
Coyle considers the work to be done by the Bibliographic Framework Transition and sees three possible scenarios for moving library data to linked data on the web: 1) go native 2) extract 3) serialize.
A serialization scenario involves switching out MARC containers for a linked data serialization. The pros to this approach is that serialization can be automated (adding identifiers to resources and data elements); however this approach results in data that continues to be siloed and lacks rich linkages to other resources on the web.
In her discussion of the extraction scenario, Coyle compares the approaches of the British Library, the National Library of Spain, and OCLC. The pros: library data can link up to data on the web. The cons to taking the extraction approach: the data isn’t visible to catalogers, making quality control difficult, local metadata will not contain identifiers, and new metadata records created after the batch extraction would have to be extracted separately.
The final scenario that Coyle discusses is the ‘go native’ option, that is, create linked data natively as a part of the daily metadata creation workflow. Pros: data is interoperable with the web, and RDA can be fully realized in a way that it won’t be with a MARC serialization. The cons: libraries would need to adopt completely new systems for handling and creating metadata without the benefit of a cost/benefit argument.
Returning back to the first step, creating identifiers, Coyle recommended the following:
- Reuse existing identifiers–don’t reinvent the wheel
- Focus on creating URLS for FRBR Group 2 and 3 entities–that is, personal and corporate names, places, and subject–as these are of benefit to the broader web community; Group 1 WEMI entities are too library-focused to do web community much good
Many in the room were asking, what’s the payoff? Why move to linked data? What new services will linked data make possible? What are the costs involved in building technical infrastructure, migrating data, developing or purchasing new metadata creation tools, and training metadata creators and managers? All of these questions need to be examined further.