Library Linked Data: e-Forum Summary

Linked Data refers to practices for publishing and linking data—cataloging data; research or government data; subject heading lists and authority files; and the riches of the World Wide Web. Using linked data methods (cataloging standards such as RDF, as well as Web protocols such as HTTP), the scope of the linked, and always further linkable, data space is incomprehensibly large—the potential is thrilling! The hope for libraries is that we can move our data into this massive data space and thereby increase the global interoperability of library data. An archive of the discussion is available at: http://lists.ala.org/wws/arc/alcts-eforum/2014-03/thrd1.html

The discussion was structured into five "scenarios" (such as having a digital collection with metadata and deciding how to convert the metadata into linked data) and a few questions to initiate discussion. The first day dealt with how to create linked data in RDF, how to store and publish this data, and how to write a schema. We shared some RDF validation tools, in the interest of reconciling text strings in a local RDF data set to text strings in a target RDF data set, to get URIs from the target set. Some specific themes that came up were: cleaning up metadata exported into CONTENTdm with Open (Google) Refine; synching linked data sources (extracting from abstracts & descriptions); and concern over URI persistence, even after RDF affords discoverability. Higher-level questions were asked, such as how to avoid silos of data even if they're linked within themselves, and how people outside of the library world find out about our identifiers or vocabularies. To help such separate data stores (especially library/non-library data stores) find each other, Schema.org's Schema Bib Extend Community Group's extensions was brought up as a possible solution. Also, RDF about RDF (RDFS, ontologies) should help, as should all of W3C's efforts to establish standards and build interoperability/compatibility should help. If we seed our data with non-library URIs, we should lead others to our data, and our users to other data.

The second day addressed challenges that we experience or expect with the transition to a linked data environment, as well as some open discussion topics. Some of those challenges relate to workflow, including authority control. Some participants pointed out that paying attention to the right fields now would enable RDA-compliant records. Controlled vocabularies enable picking up records containing URIs, and local vocabularies enhance discovery and further linking. Some techniques were shared that use XSLT to transform RDF to create linked data.

A question about addressing data sets linked to journal articles proved too new or outside the participants' purview to generate discussion.

—submitted by and e-forum moderated by Theo Gerontakos, University of Washington, Seattle; Brad Gulliford, University of Texas, Arlington; and Jeremy Myntti, J. Willard Marriott Library, University of Utah, Salt Lake City. This e-Forum was held March 11–12, 2014.