Written by Anne-Marie Viola, Metadata and Cataloging Specialist
Last week I was one of the lucky few to attend the second-ever Linked Ancient World Data Institute, along with my fellow Dumbarton Oaks colleague and resident sigillographer, Jonathan Shea. This year’s gathering was held at Drew University and convened by NYU’s Institute for the Study of the Ancient World (thanks to John Muccigrosso, Sebastian Heath and Tom Elliott!). Combining the forefront of semantic web technology with Classics, archaeology, and numismatics, this foray into digital humanities was part workshop, part conference, with dozens of awe-inspiring projects presented to demonstrate the value of linked open data.
Linked open data (LOD) is the evolution of the web from the publication of content formatted as hypertext documents to the unleashing of meaningful, contextualized information, a progression enabled by developments in encoding that afford the ability to make machine-readable “assertions.” Say what, you say? Well, rather than simply putting up webpages between which we create links, we are now capable of formulating statements by encoding these relationships between data using the properties of defined ontologies, a step closer to the idea of a more Semantic Web.
LOD also enables us to engage with our data — and others’ — in new ways, including timelines, visualizations and other types of “mash-up’s” that improve discoverability. There are a number of archival projects at work developing visualizations of community networks, such as the Social Networks in Archival Context (SNAC) project, Linked Jazz, The Crowded Page, and Yaddo Circles. What do Anne Burrows Gilchrist, Bram Stoker, Ralph Waldo Emerson and Ulysses S. Grant have in common? They all corresponded with Walt Whitman, as the visualization of dozens of archival authorities records demonstrates below.
Yet this isn’t a true visualization of LOD, because none of these records represent data out on the web – each record was deposited into a single repository that SNAC maintains. In order to be valid LOD, entities must be represented by Uniform Resource Identifiers, or URIs. These URIs, which often take the form of HTTP URLs, enable us to unambiguously define an entity by creating a pointer in the datasphere from which to begin.
For a visualization of LOD from LAWDI take the graph below. Begin with the red sphere in the center, representing Mende, the mint at this ancient site in Macedonia, identified by the URI, http://nomisma.org/id/mende (a great example of a semantically simple, or “clean” URI). Click on the URI and it resolves to an html version of a Uniform Resource Location on the Nomisma website (to learn how URIs resolve to URLs read the W3’s article on Cool URI’s for the Semantic Web).
Moving from the red sphere towards one of the black nodes you can see that Mende is “partOf” the field of Greek numismatics, find its GIS coordinates, and learn that there are related resources on Wikipedia and Pleiades, a gazetteer of ancient sites. In addition, you can discover other mints, as well as items from Mende, that are held by the American Numismatics Society (ANS). All of which is accomplished using the various vocabularies for general description and for specific disciplines, such as nomisma.org for the numismatics field.
This is the interesting part – where digital scholarship can cross disciplines to realize new knowledge. The Pelagios project that Leif Isaksen presented uses linked open data to connect various ancient world resources, such as the coin collections of ANS with the archaeological finds described in Open Context through a place name in Pleiades. As Corey Harper, Metadata Services Librarian at New York University, explained during his presentation on linked open data in the libraries, archives, and museum (LAM) fields,
Libraries have always assumed a closed world. We need to adapt to how the web changes the game.
He went on to explain that this open world is a “continual integration of heterogeneous MD [metadata], which requires self-describing data” and encouraged fellow LAWDI participants to consider the value of “a distributed bibliographic control environment.” Libraries have a long history of using both authorities and standards, but these have always been as references. LOD raises the interesting question for LAMs: what if instead of focusing on generating descriptions that stand alone in our database silos waiting for users to find them, we focus on enriching the general web and further contextualizing our scholarship by linking our unique data to others’ resources?
LAWDI helped answer that question for me. Dumbarton Oaks has materials that we can better contextualize and enhance discovery of through links to projects like Pleiades and the Virtual Identity Authorities File (VIAF), as well as the Getty vocabularies once they are released as LOD, as is expected later this year. We also produce a variety of types of data to which others could establish links, such as the unique items in our library, archival, and museum collections, as well as the creators of these materials, our publications, our staff, and – where not currently available – the terms or concepts necessary to describe all of these. As Ellen VanKeer of Brussels’ Royal Museum of Art and History proposed,
Creating LOD enables us to demonstrate relationships between the various materials within our very own institutions
such as the book held in our library and published about the collection in our museum, or the photo in our visual resource center and the archival records that document its creation as part of a fieldwork project.
Here within ICFA, we are already producing XML exports of the descriptions we craft about our archival collections and the creators of those collections, along with taxonomies of Byzantine place names and subject terms, which we use to do so. Where we have unique information to add to an entity’s identity or the historical map, we want to share this data with the world outside Dumbarton Oaks. By linking our data, we can also accomplish the reciprocal goals posed in my question above: both adding to the scholarship of the field — “data-sharing as publication” to quote Eric Kansa of Open Context — as well as to the context of our materials.