Digging into Linked Data: Perspectives from the Long Tail

The success of the semantic web depends on widespread participation by cultural heritage institutions and other organizations in making connections between open, structured datasets. Large university libraries are beginning to make such connections it s time for mid-size and smaller libraries to take the leap and establish themselves as playing a part in this web of data. In particular, digital collections of many of these libraries represent significant regional or local history collections; metadata of these collections exposed as linked data can bring visibility for these unique resources. But do these libraries have the resources to create semantic data? What kinds of resources and technical support do these libraries need? How much and what kind of training do their staff need for linked data projects?

This presentation focuses on a collaborative linked data project between two mid-sized academic libraries--Western Carolina University and Appalachian State University. The libraries are members of the Western North Carolina Library Network and share a common catalog. Both libraries have significant special collections on Appalachian culture and history. Their project aims to expose a slice of their digital collections on Appalachia as linked data and build connections to related datasets on the web thereby exploring the possibilities of the semantic web. The project also serves as the testing bed for future such collaborative work, possibly on a larger scale.

The presentation will highlight the successes and challenges faced by the presenters as they delved into this project. For example, what resources and training did they need? How successful were they in in manipulating digital collections metadata in OpenRefine; navigating the intricacies of various data models such as those from Europeana and DPLA; sorting through the multitude of controlled vocabularies that are available as linked data on the web and selecting the best possible options? How difficult or easy was it to figure out linked data jargon, such as dereferenceable URIs and RDF skeletons? What kind of technical support was needed for setting up triples stores and querying linked data via SPARQL endpoints? The presenters hope this presentation will be a useful learning experience for those who are thinking of venturing into creating access for their special collections using linked data tools particularly for those from mid-size to small libraries.