Overcoming A World of Silos
To realize its vision as a distributed-but-unified national, digital library, the DPLA will need to transcend the current barriers to interoperability across repositories of content, and the tools and interfaces used to access them.
While vast pools of digital information are currently available to library patrons—and still more might be available through a coordinated effort such as the DPLA—in the current scheme, access to these resources is typically only offered through the applications associated with that institution or repository. Scholars whose research takes them to resources provided by more than one provider will be quite familiar with the shortcomings of this situation. Discovery of resources typically works well within a pool, but poorly across different providers. Reading interfaces from different e-journal and e-book providers each have their own functionalities, interfaces, strengths and weaknesses. Amalgamating materials from different sources into a single viewing or analytic environment can be painstaking, or even impossible, depending on how local providers broker access to content.
In short, the current library world is one of silos. (See Figure 1.)
Silos have their advantages (see Figure 2),
But also their limitations (Figure 3).
While siloed presentation of content provides obvious limitations to a user’s capabilities to discover, view and interact with content, it also has fundamental implications for both those institutions that publish content to users, and for the providers of tools to interact with that content.
The fact that content delivery is typically tightly coupled to content hosting forces repositories of library materials (in this case, using “repository” in its broadest sense) to also serve as application developers and hosters. That is, if a repository holds content and wants to provide access to it, the onus is on that repository to provide discovery and access systems. And if a repository wants to provide specialized tools beyond basic access—to support annotation, manipulation, citation, personalization—this typically requires still more application development and support.
For the developers of tools that work with library information resources, the siloed approach to information delivery presents a fractured landscape of content and opportunities. While they may have developed compelling tools or environments that support rich and deep interaction with library digital resources in particular environments, their applicability is typically not portable across multiple silos of content providers. Bookmarking or highlighting features that work in one reading environment, e.g., will not work in another environment.
Digital Medieval Manuscripts – A Microcosm of Interoperability Opportunity
While the fact, and limitations, of a siloed information environment are evident across domains and pools of resources, the study of digital medieval manuscripts offers a telling microcosm of the issues and opportunities to create an interoperable network of scholarly resources.
Currently, there are a host of high-quality sites that offer access to digitized images of medieval manuscripts. Exemplifying the siloed structure of digital information from multiple providers, each has its own data store, interface, and delivery capabilities. Robust cross-site search does not exist. Medievalists seeking to study manuscripts from different repositories must contend with radically different presentations of search and viewing behaviors. Those who would seek to use specialized tools to do in depth analysis of a manuscript—annotation, mark up or transcription, e.g.—must attempt to obtain local copies of the images and their associated metadata to load into their specialized environmens (Figure 4).
To counter this environment, participants in the Digital Medieval Manuscript Interoperability Project (https://lib.stanford.edu/dmm), funded by the the Andrew W. Mellon Foundation, have proposed and built an interoperable environment for digital manuscripts that separates repositories of images and metadata from the tools and applications used to access them, with the net effect of providing scholars with an open environment in which they can discover, view and manipulate these resources using their tool of choice. (Figure 5).
This interoperable environment is specified to support:
- cross-repository discovery of resources, through metadata aggregation and indexing
- display and viewing of image-based resources from any repository in a common environment, using preferred tools
- dissemination of images and metadata to third party tools for in-depth manipulation
- persistent citation and annotation in a linked-data-friendly way
- support for differential restrictions on access to content, as may be required by license agreements
This video provides a screencast of this interoperable environment in action, with cross-repository search and browse, dynamically accessing images, descriptive and structural metadata via standard APIs in a rich page viewing environment, and then pivoting from that image into two specialized manuscript tools—one for transcription and one for annotation—that also dynamically access the manuscript content from the source repository.
Integral to this environment are two classes of API’s:
- metadata services, that syndicate the description of the structure and contents of information resources
- image delivery services, that enable access to and parameterized views of, images from source repositories in third party applications
The data model underpinning the metadata services has been specified in the SharedCanvas http://shared-canvas.org/ data model, which was heavily informed by the design work of Open Annotation Collaboration, another Andrew W. Mellon Foundation-funded initiative. A paper detailing SharedCanvas and its applicability to digital medieval manuscripts, won best paper at this year’s JCDL conference.
The image delivery API is based on djatoka’s core capabilities, but assumes neither JPEG2000’s nor djatoka as prerequisites.
More information on the Digital Manuscript Interoperability Project is available at the project website: http://www.stanford.edu/group/dmstech/, and at the prototype interoperability portal: http://dms-dev.stanford.edu.
While the DMS Interoperability project provides an instructive case study on breaking down information silos to enrich the scholarly landscape in a particular domain, the pattern and approach taken are broadly applicable. Access to image-based resources is fundamental to many research disciplines, scholarship and the transmission of cultural knowledge. Digital images are a container for much of the information content in the Web-based delivery of images, books, newspapers, manuscripts, maps, scrolls, single sheet collections, and even digital surrogates of textiles, realia and ephemera.
Many of the proponents of the DMS Interoperability Project are now launching a year-long collaboration to explore extending this approach to more classes of image-based information resources. This initiative, the International Image Interoperability Framework, holds its first meeting in September 2011. If this approach is of interest to DPLA, the project partners would welcome discussions on weaving the threads of work together.