Linked Data & RDF: New Frontiers in Metadata and Access

The last decade has seen a significant revolution. Ten years ago we were observing that the web had arrived and come of age on our PC screens. Since then the world’s entertainment and information in most languages is laid at our feet - or in our pockets. One core driver of this revolution is data: data about your friends, contacts and interests; data from Wikipedia; data embedded in the web pages of successful media and commercial organizations; data that powers search engine knowledge graphs.

The vast amounts of quality curated information in libraries and their collections is missing from this picture. Why, and what can we do to fill this library-shaped information black hole? It’s a key question for libraries of all types, one that becomes even more pertinent as the data, information and knowledge-driven web evolves over coming decades.

Linked Data is at the heart of a response to this situation. It does, after all, underpin what is happening on the web outside of libraries. Many library-linked data initiatives have been successful in their own right, but as yet these library-focused approaches have had little impact on the visibility of and access to library resources. What can we learn from the wider web’s enthusiastic adoption of these techniques?

Linked data holds enormous promise for libraries, but most efforts with linked data will be rendered ineffective if search engines cannot first harvest and index library websites and repositories. Basic search engine optimization (SEO) is a prerequisite to advanced digital library and Semantic Web applications. This presentation will cover three areas:

1. lessons learned from four years of SEO research;
2. the startlingly poor semantic identities that libraries and academic organizations have and why it matters;
3. an example of a linked data project underway at Montana State University;

Linked data offers great potential for the storage and organization of information, but how can this work in a real-world environment today? The ViAF, or Virtual Internet Authority File, is a program designed to create links between common names and titles in library catalogs internationally. Discover what the ViAF is, how it is created, and its potential for linked data in an ILS.

The University of Nevada, Las Vegas (UNLV) Linked Data Project provides a case study of the complex topic of linked open data; from emerging concept in librarianship to practical outcome. The project began with a small academic library study group created in April 2012 and comprised of professionals from various functional areas. The initial goal was to better understand linked data concepts and potential benefits to the libraries. In October 2012, after reviewing literature, attending presentations, and discussing concepts, UNLV Digital Collections designed an exploratory project. Because there was very little in the literature about how to practically implement linked data in digital collections, the team decided to focus on the transformation of typical digital collections metadata. The project made significant progress, outlining technologies, tools, and models that can be implemented by librarians. This presentation covers basic concepts of linked data, the rationale for libraries to start preparing for adopting linked data, followed by a demonstration of visualization tools operating on the linked data generated from UNLV’s digital collections. While they have not yet designed our interface, they believe that the demo will generate both insight and enthusiasm about the potential benefits of adopting linked data.

With BIBFRAME on the horizon as a replacement for the MARC 21 encoding standard, “linked data” is the current catchphrase to use when talking about library catalogs. But what does a linked data environment mean for libraries and their catalogs, especially when it comes to privacy? This session will explore the pros and cons of linked data when it comes to protecting the right to privacy.

The transition to a new bibliographic framework (now called “BIBFRAME”) is an initiative launched by the Library of Congress in 2011 to investigate and move from the current MARC-based environment to one that more fully integrates with and reaps the benefits of the World Wide Web and Linked Data. Among the goals of this new “framework” are improved discoverability through more exploitable metadata; extension of linking capabilities using URIs; the need to accommodate all types of libraries, from small to large; and, the ability to do all of this while still taking advantage of the large body of existing MARC data through transformations and reuse. This presentation will provide background information on the development of BIBFRAME; an introduction to the BIBFRAME model; examples of MARC data elements transformed to BIBFRAME statements; an overview of the BIBFRAME vocabulary; and, demonstrations of the BIBFRAME Web site and tools.

OCLC is experimenting with Schema.org, the vocabulary that is understood by the world's major search engines. This seems especially relevant now that information requests are more likely to begin with Google, Bing, Yahoo, or Yandex than in a library or a library website. Schema.org permits the creation of broadly understandable structured data, giving the library community many opportunities to enhance their visibility on the Web by creating collection points for authoritative information about the important real-world entities that populate their collections, such as creative works, authors, organizations, and publishers. But there is also a pragmatic reason for taking a look at Schema.org. It is a maturing and sophisticated ontology that allows, in words commonly echoed by Semantic Web experts, for simple descriptions to be simple, and for complex descriptions to be possible. This presentation illustrates the wisdom of this sentiment with an overview of the model of creative works derived from Schema.org, which underlies the RDF statements accessible from WorldCat.org.

When RDF was developed at the W3C, it was to meet a defined need, expressing people's privacy preferences while using the then new web. With so much present activity around libraries and RDF, semantic technology and linked data, it is imperative to understand and stay focused on the defined needs that drive developments.

In Zepheira's work with BIBFRAME, including development of the standard itself on behalf of the U.S. Library of Congress, they have always paid attention to the need to improve the visibility of libraries on the web. As a result, BIBFRAME focuses less on rigorous modeling and vocabulary and more on supporting the emergence of data patterns that take advantage of web technology to gain prominence among web users. The Libhub.org initiative founded by Zepheira is an effort to establish a BIBFRAME-based linking network towards making library resources as near at hand to students, teachers, researchers, policy-makers and casual browsers as sites such as Wikipedia.

In this presentation, Uche Ogbuji will lay out key design elements of BIBFRAME in practical context, touching on topics such as RDF and schema.org. He will report on the in-progress efforts by Zepheira and its partners, using data from participating libraries, to apply linked data in increasing libraries' visibility and impact.

The Tutu Ontology (TUO) and Measurement Ontology for Tutus (MOT) have been created to address the absence of a domain-specific formal specification for terms and relationships for representing dance artifacts. These ontologies meet the specific need of the dance community to have access to archival resources and engage in the archiving process. Used in combination, the ontologies allow dance organizations and individuals to explore and share information about dance costumes and to track or locate existing costumes. The TUO ontology provides elements that align specifically to the vocabulary and descriptive needs of costumers and costume archivists for representation, historical documentation and collection management of tutus. The complementary MOT ontology provides elements for specifying the physical dimensions of a costume or a dancer with a vocabulary appropriate for pattern design, construction and body fit. Formalized using RDF/RDFS and OWL, TUO and MOT work together to give users the ability to create archival records for dance costumes, manage collections of costumes in use and in storage, share collection information though semantic web technologies, and provide the ability for exploration and presentation of the history of an individual costume or costume collection across time.

The SPARQL Protocol and RDF Query Language is an essential linked data technology used to create, maintain, and retrieve RDF information. This session will provide an introduction to basic SPARQL query syntax including SELECT, FILTER, UNION, and the use of variables. Attendees will learn how to write simple SPARQL queries that reveal what kinds of data a triple store holds, and how to develop triple patterns that retrieve data in useful ways. The presentation will demonstrate the use of the Apache Jena ARQ tool to query local datasets and online SPARQL endpoints, and provide pointers to online resources for practicing SPARQL queries and working with linked data.