XI. Survey of Current Library Linked Data Implementation

Below is a representative Survey of national and research libraries currently engaged in Linked Data implement or Experimentation. It is by no means a comprehensive list. Rather, it is meant to serve as an indicator of the various states of adoption and readiness.

Library of Congress:

Library of Congress has initiated several Linked Data projects since 2009. It created its Linked Data Service by publishing its authority data and other vocabularies in Linked Data format (id.log.gov) and developed BIBFRAME model. In late 2015, LC launched its BIBFRAME pilot project which was designed to test “efficacy of BIBFRAME.” More than 40 LC catalogers participated in Phase One of the pilot (Oct. 2015-March 2016). The report and assessment of the pilot project can be found at: https://www.loc.gov/bibframe/docs/pdf/bibframe-pilot-phase1-analysis.pdf. LC is planning to launch Phase Two in early 2017 which will experiment with BIBFRAME vocabulary version 2.0.

Library of Congress has created three sets of BIBFRAME tools which libraries can use for their own experiments:

1. BIBFRAME Editor: Library of Congress BIBFRAME Pilot Training for Catalogers Module 3 Unit 2 (http://www.loc.gov/catworkshop/bibframe/Module3Unit2External.pdf) provides detailed instructions on how to use the Editor.
2. BIBFRAME Profile Editor: The document, BIBFRAME Profiles: Introduction and Specification (http://www.loc.gov/bibframe/docs/bibframe-profiles.html), defines what BIBFRAME Profiles are and describes how they are constructed.
3. MARC to BIBFRAME transformation tools: The software that underpins these two services can be downloaded from GitHub site: https://github.com/lcnetdev/marc2bibframe.
National Library of Medicine:

The National Library of Medicine was one of the BIBFRAME Early Experimenters and a registered BIBFRAME Early Implementer. In late 2014, NLM proposed a modular approach to BIBFRAME (BF) experimentation through development of a core ontology, i.e., a widely shareable BF vocabulary, that could be extended with other RDF ontologies for greater granularity. To test this approach, NLM collaborated with Zepheira, George Washington University, and University of California, Davis (UCD) in the early development of the BIBFRAME Lite (BF Lite) ontology suite (http://bibfra.me/).
In addition to its early work on BIBFRAME and BIBFRAME Lite, in 2014 NLM published beta versions of two of its well known datasets as Linked Data: PubChemRDF, containing information on the biological activities of small molecules (https://pubchem.ncbi.nlm.nih.gov/rdf/) and MeSH RDF, NLM’s thesaurus of Medical Subject Headings (https://id.nlm.nih.gov/mesh/). Both RDF products are searchable from their own SPARQL query interfaces or querying can be directly integrated into programs and services using their SPARQL endpoints.

George Washington University Libraries:

The Linked Data experiment George Washington University Libraries (GW) has undertaken is to insert URIs into MARC records by utilizing MARCNext Linked Identifiers of the MarcEdit open source Tookit. To date GW entered over four million URIs into its existing legacy MARC records. Internal cataloging workflow was adjusted to accommodate and ensure the newly added MARC records will contain appropriate URIs. The embedding of URIs via MARCNext is automatically set to Library of Congress ID service (id.log.gov) as the default setting. Additional vocabularies, e.g. VIAF, other national libraries, and OCLC are also possible. The URIs are embedded in $0 of the following MARC fields: 1xx, 240, 6xx, 7xx, and 830. GW’s URI project is an excellent example demonstrating how libraries can apply Linked Data concepts within their existing MARC-centric systems as a transition to a data model that is more Linked Data friendly.

Linked Data for Libraries and for Production Projects:

Linked Data for Libraries (LD4L) is a series of two integrated projects (LD4L and LD4L-Labs) supported by grants from the Andrew W. Mellon Foundation (http://www.ld4l.org/ld4l-original/). They involve several major research libraries, including Cornell, Harvard, Stanford, and the University of Iowa. Those projects were designed to examine and test the discovery of Linked Data (LD4L) and to create tools and services that support the creation of Linked Data (LD4L-Labs). The ultimate goal of the LD4L projects is to create Linked Data solutions and infrastructures that can be implemented in a production environment at research libraries within the next three to five years.
Linked Data for Production (LD4P) is a related project also funded by the Mellon Foundation. This project involves Columbia, Cornell, Harvard, the Library of Congress, Princeton, and Stanford. The goal of LD4P is to begin the transition to the native creation of Linked Data in a library’s current production environment using existing tools. Issues LD4P discovers with current tools will be fed back to LD4L-Labs to aid them in their future tool development.

The British Library:

According to Neil Wilson, Head of Collection Metadata at the British Library:

The British Library believes Linked Open Data to be a logical evolutionary step for the established library principle of freedom of access to information. As such it offers trusted and authoritative knowledge organizations such as libraries a new important role in the emerging information landscape. The vision of a global pool of semantically rich, reusable metadata is also highly attractive to such organizations by enabling the concentration of scarce resources on adding unique value. Similarly, the potential of Linked Data for cost effective exposure of library datasets to search engines, application developers and new forms of resource discovery has significant appeal.

Given the above commitment, it is not surprising that the British Library was an early adopter of Linked Data technologies. In 2011 they released metadata for a subset of the British National Bibliography (BNB) under a Creative Commons CC0 1.0 Universal Public Domain license. Metadata from the bibliography is made available via a SPARQL endpoint (bnb.data.bl.uk) in addition to downloadable data dumps serialized as RDF/XML and N-Triples.

Data in the BNB dataset is described using the following vocabularies:
• Bibliographic Ontology
• Bio
• British Library Terms
• Dublin Core
• Event Ontology
• FOAF
• ISBD
• MADS/RDF
• Org
• OWL
• RDA
• SKOS
• WGS84 Geo Positioning

Explicit URI based linking in the metadata is established with the following data sources:
• ISNI
• VIAF
• LCSH
• Lexvo
• GeoNames
• MARC Country and Language codes
• Dewey.info
• RDF Book Mashup

The Library’s Linked Data model gives preference to the use of pre-existing ontologies in order to ensure the widest options for interoperability with new user groups in addition to the library world. Since its creation, the Linked Open BNB has continued to be enhanced through the addition of new features such as the ISNI and regular monthly updates. The British Library is also looking at new options to expand its Linked Data services including the use of content negotiation for selected areas of its web site to offer Turtle (TTL) versions of pages in addition to standard HTML.
Although The British Library was a member of the BIBFRAME early experimenters group and remains committed to offering linked open data, it currently has no plans to adopt BIBFRAME as the framework for doing so. This does not mean the British Library is uninterested in BIBFRAME; but simply reflects that the organisation is waiting for BIBFRAME to achieve the necessary maturity, stability and critical mass of users in order to justify the commitment of limited development resources.

University of California, Davis:

The research contained in this report is a reflection of the current state of the UC Davis Library’s engagement with Linked Data. The UC Davis Library plans to continue its efforts on this front by moving to implement the transition plan suggested in this report as part of its regular operations. To this end, a Linked Data transition has been formed, and we expect to begin an implementation process during the Summer of 2017. As we engage in this transition, we will continue to update this roadmap to document the transition process and communicate specific, relevant findings. Changes will appear in versions made available at bibflow.library.ucdavis.edu.

<<  Discovery Appendix A: Vendor Engagement Matrix  >>
Return to BIBFLOW Roadmap Table of Contents