Skip to Main Content

Penn Libraries Linked Data Framework: Near Term Framework for Linked Data

3. Near Term Framework for Linked Data

The near-term framework for linked data at Penn Libraries involves building staff skills and expertise in linked data; researching effective workflows, tools, and collaborations for linked data management; and enhancing and extending existing data models and systems using hybrid linked data to improve discovery.  Aspects of this hybrid approach include introducing linked data alongside existing data structures like MARC.


Library staff have worked on several projects to create and manage linked data and improve their skills in it. In collaboration with the Program for Cooperative Cataloging (PCC), staff have enhanced Wikidata with information on serials and on Penn departments, to support our Deep Backfile copyright project and to lay the groundwork for collecting information on publications and other work of Penn scholars.  Staff have also trained, experimented with, and consulted on the design of Sinopia for describing bibliographic resources with linked data.  The Digital Scriptorium 2.0 project plans to maintain information on library manuscript holdings using linked data on a Wikibase instance.  Along with manually editing linked data, several library staff also have used automated or semi-automated tools to create, manage, and retrieve linked data, including OpenRefine, QuickStatements, SPARQL queries, and RDF-generating scripts.


The “Possible Linked Data Connections at Penn” diagram shows preliminary ideas we have on how linked data workflows might work at Penn.  
















Possible augmentations of discovery functionality using linked data include the following:


BibCards retrieve and assemble knowledge card information about the authors found in bibliographic data, to provide context for search results. BibCard implementations use identifiers like Library of Congress Name Authority File (LCNAF) as input, and crawl linked open data sources.  At Penn they could also draw on data from Share-VDE’s hosted triple store. User studies show that these potential changes are evaluated favorably (see, e.g., 


Type-ahead and entity suggestions provides search assistance to catalog users. Search assistance can dynamically suggest alternative terms, query reformulations, and possibly “best bets,” search suggestions in a discovery layer. This functionality can draw on Share-VDE APIs within the discovery environment, using Share-VDE infrastructure to dynamically pull in search assistance data.