As more institutions use linked data, more possibilities arise for cross-organizational collaborations in sharing linked data, and in co-developing tools for managing linked data. Lux, a linked data discovery platform developed at Yale, provides one example of such a collaboration. Lux imports metadata from Yale’s libraries and museums, as well as from widely used vocabularies and thesauri from institutions like the Getty and the Library of Congress, as well as Wikidata, to produce a linked data graph of millions of entities that users can search and browse to discover resources in different Yale museums and libraries, and see how they relate to each other, and how they relate to common places, topics, and events. Lux’s linked data combines and normalizes metadata already produced in participant libraries and museums, who do not have to change how they manage their own metadata. The Lux developers plan to release their software as open source and invite other institutions to use it as well. Cross-institutional use and coordination of systems like Lux could enable users to explore seamlessly across multiple institutions’ collections available to them (as Yale’s is to Penn library users through programs like BorrowDirect).
While the linked data in Lux is derived from managed data, other collaborative linked data projects manage linked data directly. The Blue Core project, which the Penn Libraries are currently planning with Stanford, Cornell, and the Library of Congress, proposes that libraries manage their bibliographic data in a shared BIBFRAME linked open data store. The Blue Core data repository would support import from MARC and export to MARC, but the metadata would be maintained as BIBFRAME data. Such a shared, open data environment could make it easier for libraries to maintain bibliographic catalogs, and to extend their functionality, than current systems that require copy cataloging or that have proprietary encumbrances.
Wikipedia’s experience with linked data suggests an evolutionary path like that suggested by adopting systems like Lux and Blue Core. Wikipedia’s first major linked data project was DBPedia, which consisted of linked data automatically generated from Wikipedia articles that were manually edited and that contained structured data but not explicit linked data. DBPedia was depicted at the center of the early linked data cloud, but in recent years it largely has been superseded by Wikidata, in which people and programs directly create and update native linked data, which can then be included both in Wikipedia articles and in many other applications. DBPedia was the easier application to create initially, with its linked data downstream from where data was maintained, but Wikidata has shown itself over time to be more timely, extendable, and applicable to a variety of downstream uses. Similarly, “downstream” linked data applications like Lux may be more feasible for libraries like Penn to deploy initially, but native linked data systems like Blue Core may have more potential in the long run, even if they are more difficult to create initially.
