Many academic libraries provide services to support students and faculty in the submission and archiving of electronic theses and dissertations (ETD). In the state of Texas, the Texas Digital Library (TDL) is a consortium of eighteen universities and has a mission of providing common infrastructure, services, and training to support the scholarly communication needs of its member institutions.¹ Among other services, TDL provides platforms for hosting open-access journals and wikis, and it supports a federation of institutional repositories.

These institutional repositories are running the DSpace software. Some members host their own DSpace repository; some share a repository at another member's institution; and others use a shared repository hosted by TDL. These institutional repositories provide a publishing and archiving platform, typically for born-digital documents such as journal article preprints, conference papers, and technical reports. Some members were using their institutional repository for the submission of theses and dissertations, too.

Vireo

http://www.tdl.org/etds

With support from the Institute of Museum and Library Services, TDL began a process of seeking input from its stakeholders to design a new system for managing and preserving ETDs. The project, named Vireo, sought to leverage existing infrastructure, implement new workflows, and scale up to a distributed, statewide ETD system. The Manakin software was used to create a customized user interface for DSpace to enable students to submit their dissertations.² The dissertation and its related files and metadata are then stored in the local DSpace repository.

At a high level, DSpace repositories are organized into communities, collections, and items. Items are made up of metadata and bundles, which contain one or more bitstreams.³ For example, a university department may constitute a community, which has a collection of technical reports. Each individual technical report may be represented as an item that includes a metadata record that describes the report and a bundle of files, which can be thought of as bitstreams. In this case, there may be a bitstream that represents an Adobe Acrobat file (PDF) of the report.

In mapping the DSpace data model to ORE, TDL decided to define aggregations for communities, collections, and items. It wrote some code to enable each DSpace repository to generate and interpret Resource Maps for these kinds of aggregations and to expose them as metadata records using DSpace's OAI-PMH interface. (Revisit chapter 1 for more information about the OAI-PMH.) The Resource Maps are serialized as Atom XML and can be harvested by an OAI-PMH service provider by specifying the proper metadata prefix.⁴

For its central DSpace repository, TDL employed an OAI-PMH harvester to harvest the Resource Maps and developed an ORE item importer. The ORE item importer resolves the URIs of the Aggregated Resources described in the harvested Resource Map, fetches them from the Remote DSpace repository, and rebuilds the item with its bitstreams and metadata in the central DSpace repository. TDL also built a custom scheduling system to automate harvesting.⁵

In this way, TDL is using ORE to harvest all of the dissertations and metadata from its member institutions into a central repository where they can be more easily preserved and made accessible in a single location. Future plans include public syndication of the Resource Maps so that anyone on the Internet can access and use the ETDs in semantic applications. In fact, interest has already been expressed by water quality researchers in Texas who want to automate the harvest of data from dissertations that relate to their field.

TDL invested a great deal of time in developing and testing its software because it is implementing the software in a production environment with thousands of users. It is planning to release its ORE modifications to DSpace as open source software in February 2010.⁶

Foresite began as a project funded by the Joint Information Systems Committee (JISC) in the United Kingdom to produce a demonstration of the ORE standard by creating Resource Maps of journals and their contents from the JSTOR archive of academic journals and delivering them as ATOM documents to deposit in a DSpace repository using SWORD. The Resource Maps were ingested into DSpace as items that reference the content residing in JSTOR.⁷

Foresite libraries

http://code.google.com/p/foresite-toolkit/

Foresite is probably more commonly known for producing open source Java and Python libraries for constructing, parsing, manipulating, and serializing Resource Maps. Both sets of libraries support the parsing and serialization of Resource Maps that are suggested in the ORE specification: Atom XML, RDF/XML, and RDFa. Additionally, they support serialization in Notation3 (N3), N-Triples, and Terse RDF Triple Language (Turtle).

We wanted to operate at the object-level, with a protocol that is focused on objects and not just the metadata. The intent with Vireo was not to produce something that can harvest metadata but something that can harvest metadata and items inside of collections … and that's one of the things that was exciting to us about ORE. We want to share everything we can possible share in order to maximize the scholarship contained in the Texas Digital Library.

—Mark McFarland, Texas Digital Library

ORE is used for describing compound/complex digital objects such as aggregations of journals, issues, articles, and pages within JSTOR and enabling the digital preservation all of the copies of a resource. Of the two sets of libraries, Foresite's implementation of ORE is more complete in Python than Java. In the Python libraries, Foresite hides the ORE data model (in RDF) underneath an object-oriented layer and familiar “pythonic” style. It was used to create ORE descriptions of the complete holdings of JSTOR, making available the graph of interconnected journals, issues, and articles, through structure as well as citations.⁸

JSTOR is currently modifying the Foresite code to use its own internal formats rather than the information exported in the original project. It hopes to make the Resource Maps available to users at some point in the future.⁹ The Foresite libraries are available for download from Google Code.

Microsoft External Research partners with universities to support research, traditionally in computer science, but also in other areas such as library science and e-Science. Along with supporting research projects that are directed outside of Microsoft, External Research engages in activities such as sponsoring academic conferences, providing fellowships and internships, and producing software tools to foster and improve the research process. Microsoft provided early support for the development of the ORE specification along with the National Science Foundation, the Andrew W. Mellon Foundation, and the Coalition for Networked Information.¹⁰

The most important thing to know about ORE is not ORE at all, but the web architecture on which it is based. Understanding how the web works, and then how RDF works, is more important than knowing the details of ORE. Once the fundamentals are understood, ORE is very straightforward. Understanding that it is not just another compound object format like METS or MPEG21 DIDL is the most important thing for libraries. It could be used for Archives in place of Encoded Archival Description (EAD), for compound objects in place of METS, for collections in place of proprietary databases or description formats.

—Robert Sanderson, Los Alamos National Labs

The goal of the U. S. National Virtual Observatory (NVO) is to “enable a new way of doing astronomy” by to making it possible for researchers to find, retrieve, and analyze astronomical data from ground- and space-based telescopes worldwide.¹⁸ The NVO is sponsored by the National Science Foundation and is based at Johns Hopkins University (JHU). Librarians at the university's Digital Research and Curation Center were early collaborators with astronomers in building solutions for submitting, publishing, and curating data sets for the NVO community, applying many of the principles of library science to the management of large, astronomical data collections. Tim DiLauro, Digital Library Architect at the JHU Sheridan Libraries, describes their project:

The overall goal of the project is to capture the data that is associated with publications, deposit them into a data archive, and enable services over the data in the archive. One of the most fundamental aspects of scientific scholarly communication is the ability to access and examine cited data. Without this ability, the very essence of the scientific method, with its requirement of validating results, becomes compromised. The NVO is playing a leadership role in building services for the astronomy community to access and analyze astronomical data. However, thus far the scope of the NVO has deliberately not included long-term data curation, focusing instead on data location and data access standards and protocols. One of the goals of our project, which is a collaboration of astronomers, a scholarly society, its publishing production partner, and research libraries, is to capture data that is related to a journal article when it is submitted [and archive it]. The challenges are several:

• To gather more metadata and datasets from authors without significantly increasing their workload,
• To simplify deposit process for authors and publishers, and
• To enable linking between articles and datasets without significant impact on publisher systems.

To accomplish these goals, we chose ORE as an enabling technology.¹⁹ In the project, they are using ORE to support the description of the relationships between data and an article. For example, an article may include images, tables, and graphs that are embedded in it. Treating the article as an Aggregation, a Resource Map is generated that identifies and links the data behind these embedded objects and the article when it is submitted. If the article was written using Microsoft Word for Windows, the Resource Map can be created by the Article Authoring Add-In. JHU has also created a web-based application that can generate a Resource Map for other formats that are common in Astronomy, such as LaTeX. In either case, the document is submitted with its Resource Map using SWORD to the publishers.²⁰

Unlike the other ORE implementations described in this chapter, JHU does not maintain the Resource Maps after they are generated. They are used by the publisher to link and ingest the article and its data when they are submitted. The publishers' systems can then track the relationships in their own way.²¹

JHU considered other options, such as METS, and specifically structMaps, but decided that ORE was a better fit because it was designed to express relationships among resources and to support the expression of complex objects. They also anticipate that the tools that will be developed for ORE will align more closely with their needs in the future.²²

Along with DiLauro, who served on the ORE Technical Committee, Sayeed Choudhury from JHU's Sheridan Libraries helped develop the ORE specification by serving on the ORE Advisory Committee. The NVO is currently being operationalized by NASA as the U. S. Virtual Astronomical Observatory. JHU's implementation of ORE will rolled into the Data Conservancy,²³ one of two DataNet projects that was funded by the NSF in 2009.

The Australian Literature Resource (also known as AustLit) is a collaboration between the National Library of Australia and twelve Australian universities to index and provide authoritative information on more than 100,000 Australian authors, going back to 1780.²⁴ Literature Object Re-use and Exchange (LORE) was created in this context as a lightweight tool to enable researchers to create and publish ORE-compliant literary objects that encapsulate their digital resources and bibliographic metadata.²⁵ LORE runs as a plug-in to the Mozilla Firefox web browser. It provides a graphical tool for drawing and labeling typed relationships between objects using terms from a bibliographic ontology. Metadata can be attached to the object, which can then be published as an RDF graph to a repository, where it can be searched, downloaded, edited, and reused by others.²⁶ It stores and queries Named Graphs that represent literary compound objects using web services on a Sesame 2 (RDF triple-store) or Fedora repository. The types for relationships among or between Aggregations and metadata that describe the Aggregated Resources are specified by an OWL ontology that was developed after examining the topic types and relationships present in AustLit. The ontology is based on FRBR but was extended to support additional relationships.

The LORE authoring interface displays a graphical visualization of the Resource Maps with their Aggregated Resources represented as nodes with arrows between them representing their typed relationships. A node presents a preview of its resource such as a thumbnail image, making it easy to locate and identify resources. Clicking on an identifier will load the resource in the web browser window. Along with the visualization, the Resource Maps are displayed as RDF/XML in a text window. New resources can be added as they are browsed in the browser window. A toolbar allows objects to be saved and loaded from the repository, and another panel enables the user to search and browse Resource Maps. Finally, metadata can be added or edited in the Properties panel.²⁷

LORE was created as a part of the Aus-e-Lit project by the eResearch group at the University of Queensland²⁸ under the direction of Jane Hunter, who also served on the ORE Advisory Committee.