This is the first guest post in our blog series highlighting the work of Scholars participating in Networking Archaeological Data and Communities: An NEH Institute for Advanced Topics in the Digital Humanities (NADAC). Below, NADAC Scholar Greta Hawes shares the journey of how her project, MANTO, transformed from idea to reality. Keep an eye on this page for more guest posts in the new year!
MANTO is a born-digital initiative to map the storyworld of Greek myth and its impacts on the historical landscape of the Mediterranean.
When Scott Smith and I began MANTO in late 2017, neither of us had much experience with digital tools, but we knew we wanted to create something that was both useful to a broad audience and that showed Greek myth in a new light. We had three strokes of luck in those early months. The first was discovering through conversations with digital humanists that there was such a thing as “Linked Open Data” (LOD). The second was my stumbling across the data platform Nodegoat, which has allowed us–a team of technical lightweights–to create highly structured networks. And the third was the fact that ancient world studies already had a long history (relatively speaking) in creating collective digital resources, so we had a good community to work within.
I’ve also had the good fortune of participating in Networking Archaeological Data and Communities (NADAC), an Institute for Advanced Topics in the Digital Humanities facilitated by The Alexandria Archive Institute/Open Context and Funded by the National Endowment for the Humanities, which has taught me an enormous amount about how we can work with archaeological data in sustainable and ethical ways. In this blog post, I’m taking some time to think through how and why LOD has become important to how MANTO functions.
From the point of view of a single project, LOD principles are pretty simple: you carefully identify and disambiguate the things you’re interested in, assign each a unique code (i.e., a URI) and then you construct a system for expressing the connections between them. But it is when we step outside of a single project and explore the broader digital ecosystem that LOD really comes into its own.
LOD allows for the alignment of “things” between different projects. It recognises that no one dataset will ever be large and rich enough for all possible uses, and that, individually, every dataset is vulnerable to obsolescence. LOD sees a solution in decentralised collaboration: by linking and aggregating across datasets we can create a more sustainable network whereby each project both benefits from and expands the whole in its own way.
When we began MANTO we had the great advantage of being able to use Pleiades, a gazetteer of ancient places, which is deservedly recognised as one of the great success stories of digital ancient world studies (Fig. 1). When we align our places to the places in Pleiades, we benefit from the geographical expertise of Pleiades’ editors, and from those before them who created the resources on which Pleiades is based. Pleiades’ URIs allow us to mitigate against potential confusion by disambiguating the places we want to capture (there were, for example, some two dozen cities called “Heracleia” in antiquity…). It also allows us to easily populate our own maps by fetching their latitude and longitudinal data. In short, we don’t have to re-do for ourselves work that has already been done by others, with more expertise in this field.
MANTO’s interest in ancient places is highly specific: we want to capture the myths that were told about these places and the mythic relics that were preserved in them. So the data that we collect about places is naturally quite different from that which Pleiades holds:
We are currently working with Pleiades to add MANTO’s URIs to its existing dataset, and to add to it the kinds of places that MANTO specialises in uncovering: fictional locations like the Underworld, and landmarks like heroic tombs mentioned in ancient texts.
By contrast with places, which were pretty well catered for in LOD resources, when we began MANTO there was no good dataset of mythical heroes, monsters and gods that we could easily (re)use. Over time, we have built up our own resource, consisting now of more than 4800 “agents” (i.e., individual “people”), almost 300 “collectives” (groups of people) and almost 350 objects like weapons and clothing used by them.
Meanwhile, the digital ecosystem has matured as well. Good resources for mythic people are available at Topostext and Mythoskop. Pelagios also now has a “people” activity that is working to “to survey practice and collect guidance in the handling of interoperable Linked Open Data relating to historical person datasets”. An impressive amount of work has been done in growing this material in Wikidata, such that this is now emerging as the best way of aligning heterogeneous collections of names across datasets.
This work of disambiguating and aligning names is already bearing fruit. So, we have been able to easily ingest open access data about sealings from Tel Kedesh because the gods and heroes depicted on some of these are aligned to MANTO entities when published in Open Context. Equally, we hope to start work on integrating material from ancient epigraphy, where a number of projects (e.g., IRCyr) are using Wikidata entities to identify gods and heroes.
By far our biggest challenge, however, has been in working with ancient artifacts as LOD. Until the middle of this year, all of MANTO’s data was captured from ancient texts. Greek and Latin texts shifted online surprisingly easily: they come with few rights issues that would preclude open access, and they were already–thanks to centuries of scholarship–a well organised corpus. Projects like Perseus have created indispensable resources, and their Scaife Viewer now includes Canonical Text Services (CTS) URNs for identifying specific passages in specific texts which we align MANTO’s passages to where possible.
Capturing ancient artifacts that depict Greek myth is a much more difficult proposition. They are by nature a highly diverse group of objects–coins, painted vases, sculpted reliefs, wall paintings, etc.–and they are held in myriad collections and published in various ways. MANTO is essentially an aggregator of information about artifacts, so we rely on data that we have collected and organised from others.
Through the course of this year we have recognised that we need to both identify and classify each artifact that we have in MANTO, and we need to be able to do this in as efficient a way as possible for each individual artifact.
At this point, it’s useful to look at a specific example. Here is an Attic red-figure hydria in the University of Sydney’s Nicholson Collection that we added recently (Fig. 2):
This hydria is represented in MANTO by this “Filecard” (Fig. 3):
The fields towards the bottom of the card all represent different classifications for this artifact. With a potential dataset of artifacts in the tens (perhaps hundreds) of thousands, these classifications are crucial. They will ultimately allow us to group them so that we can filter to just see (e.g.) depictions on Roman wall-paintings, or sculptures found in north Africa, or objects made in Attica in the fourth century BCE. We reused the places in our Pleiades-aligned dataset to model where artifacts were made or excavated. To model when they were made, we decided on a system of time periods that was granular enough for our purposes while recognising that few ancient artifacts can be dated with precision. Our time periods are aligned to ChronOntology, with a few tweaks made to suit our purposes (Fig. 4). (So, for example, our “Early Imperial” period ends in 79 CE rather than 68 CE so we can use it for all artifacts buried in the Vesuvian eruption.)
Our classifications could of course go much further. So, although we capture information about the kind of vessel (amphora, hydria, etc.) and its attribution to a particular ancient artist in the name of each artifact, we don’t formally capture these aspects as LOD. But that is not to say that these could not later be added if we–or someone else–had the resources to do it. Kerameikos offers a classification scheme for ancient ceramics, as Nomisma does for coins.
The identification of our artifacts has proved quite difficult. Because MANTO draws this data from many different sources, there is always the risk that we will accidentally duplicate artifacts if we cannot easily check whether we have already added that one or not. There is– unsurprisingly–no single LOD resource that covers all ancient artifacts. Our solution, then, is to bring together as many stable identifiers as we can, focusing on the most-used resources.
So, for the Attic hydria we align it with both analogue and digital identifiers. These include its catalogue number in the Nicholson collection, and the vase’s number in Beazley Archive Pottery Database and in the original LIMC volumes. (We would also include references to the Corpus Vasorum Antiquorum if it appeared in one, as well as any inventory numbers it had had in the past.) The link to the Nicholson collection record takes us to some online information. Stable urls are provided by digital resources, the Digital LIMC, and the Beazley Archive Pottery Database. We also link to Wikidata where the artifact has been added there.
Aligning these identifiers, or at least checking the alignments that we have found elsewhere, requires a commitment of time and expertise. Now that we have the data structure established, we are working to find ways to speed up our data collection methods.
This work of developing an ontology for identifying and classifying artifacts for use in MANTO has been a collaboration with Ewan Coopey.
My thanks go to Elton Barker, Anna Foka, Katrina Grant, Terhi Nurmikko-Fuller, and Ben Swift, for instructive discussions; to Nodegoat’s developers Pim van Bree and Geert Kessels for their support; and to Tom Elliott, Jonathan Groß, Brady Kiesling, and Rosemary Selth who have done so much to create links out beyond MANTO. Candace Richards and Melanie Pitkin provided welcome guidance in accessing the Nicholson collection.
I also thank core faculty and my fellow participants at the recent NADAC summer workshop in Berkeley, for their extraordinary support and feedback on this part of the project; for their third-hand accounts of appendectomies on remote digs done with nothing more than some local moonshine, a properly sharpened trowel and a vague understanding of human anatomy; and for not being too impolite about my controversial contention that all vessels are simply vessels.