Nomisma.org is a collaborative project to provide stable digital representations of numismatic concepts according to the principles of Linked Open Data. These take the form of http URIs that also provide access to machine–readable information about those concepts, along with links to other resources. We have also constructed an ontology for representing concepts in our thesaurus, and this has been applied to digital representations of physical specimens, enabling linking between specimens in Nomisma–defined numismatic concepts.
At CAA 2016 in Oslo, we presented our experiences in setting up the Nomisma.org ontology and how to combine it with other LOD approaches [1]. Within the consortium of numismatists the usage of Nomisma.org has further increased since then, with more and more systems and users, and thus increasing expectations. For several years we have been holding an annual meeting of colleagues interested in Nomisma.org together with an interest group of numismatists working on coin finds, the European Coin Find Network (ECFN). In May 2019 the 8th Joint Meeting of Nomisma.org and ECFN was held in Messina with some 50 participants [2].
With the growing success of Nomisma.org, we are encountering new problems, while known problems are getting more and more important. The work we do can be split in two areas and issues will explained here according to these areas:
- Creating and defining concepts (IDs) like lists for material, denominations, mints, etc. that are needed for the different fields of numismatics. This task is mainly done by domain experts either via a front-end, where single concepts can be entered, or via spreadsheets that are then imported.
- The maintenance of the ontology itself.
Issues with 1):
- We try to cover different fields of numismatics, like Greek, Roman, Medieval. While Nomisma.org initially concentrated mainly on Roman, we are now expanding. The new fields can have very different material and situations, and therefore have different needs. With the growing number of existing concepts, there is a need to work more systematically and precisely, and to attach indications of the relevant fields of numismatic for each of them. Our goal is that the IDs for the concepts have a name that explains the concept, and is not just a number. This is getting more and more complex, and solutions we generate now will not always fit to those IDs that already have been created.
- We generated concepts that are used by the numismatic community, even in cases where we knew that there might be contradictions about their usefulness or interpretation. Something we struggled for a very long time was the question of whether a mint represents just a place (e.g. a town), or does it represent a single workshop that might be provided with geo-positioning. In the latter case, where it is simply the location, then the mints of Istanbul and Constantinople would simply be the same concept. In the former case, there is an Ottoman and a Byzantine mint. As a result, we took the decision to generate all of the different concepts, but to link them together semantically using prov:alternateOf from the W3C Provenance ontology [3].
Issues with 2):
- We set up the ontology without range and domain definitions in order to allow flexible usage. Especially for properties like nmo:hasStartDate or nmo:hasEndDate, this was very helpful, since they can be reused in various situations. Changes in the ontology are much less frequent. The main tasks will be to add the new classes and properties required by the growing expansion of the domain. Nomisma.org follows a bottom-up approach, both through the flexible ontology without range and domain definitions, and by involving the domain experts in ID creation. However, as already reported in [1], this results in different modelling approaches that then need to be mapped before they can be merged into one unified view.
- Since we are also increasing access to the ontology for new participants, they come with new good/valid arguments for changing some things. Or they simply spot errors we have made. Most of these issues can be handled by simply fine-tuning the definitions, but deeper changes need time and discussion; and we must deprecate and redirect to successor elements in case elements need to be removed. However, each version of our ontology should remain available via the Nomisma.org website.
Of significance for the purposes of this workshop are the following sub-topics and lessons we have learned (and continue to learn) as we address the following challenges:
- The application and extension of an existing standard, namely CIDOC-CRM, to solve the challenge of digital representation (as described above).
- The application of existing controlled vocabularies, namely the Getty Vocabularies, in the migration from a legacy database structure to Linked Open Data. While a significant component of the data is inherently commensurate with the Getty Vocabularies, some of it is not (thus far). Should the Getty Vocabularies, a resource for the broader field, be adjusted for the needs of this project; that is, would expanding, for example, the Union List of Artist’s Names to include the names of buyers and sellers be of benefit to the field or distort the nature of this resource? Who has the moral authority to assert the identity (unique resource identifier), for example, of an individual named in an inventory—the institution curating the inventory into a data resource or the institution that holds the original archival document? If we agree that it is the latter, then we must anticipate an ongoing process of identification reconciliations and concordances as many cultural heritage institutions have not yet moved into Linked Open Data. The latter would be a resource-intensive effort; are institutions prepared for this long-term approach to data sustainability, interoperability, and usefulness?
- Ontology re-use, improvement and precision. The Getty as an institution anticipates uses of the Linked Art data model for not only the Provenance Index but other digital data sources, such as the collection of digitized special collections artifacts. Yet each original archival resource has the potential to impact the data model. Even in the case of a single dealer’s business records, record keeping strategies and the information collected changed over time. What impulse/business need drives the data model: a desire for stability and long-term re-use, which may require dispensing with certain levels of granularity (and the archival evidence represented), or fully accurate representation of the art world as traced in archival and other institutional records? And, of course, full accuracy raises the specter of Jorge Luis Borges’ short story “On Exactitude in Science” which points up the impossibility of the ‘perfect’ map, which would end up representing the world on a one-to-one scale. Such precision undermines the effort of modeling the data and also thwarts effective search. This gives rise to the subtopic of …
- Tensions between simplicity and complexity or nuance.
Other issues:
- Our next steps are to continue to expand Nomisma.org, in particular in order to optimise linking to data from archaeological excavations. In order to further this, a conference “Coins in Context” was held in Oxford last year [4].
- Still open is the question of how to model/handle uncertainty? There are already some uncertain mints or uncertain geopositions included in the Nomisma ID namespace. However, we think this is a subject that needs to be solved on a larger scale. Already in 2014 at the CAA in Paris [5] we proposed some modelling ideas, while this year at CAA in Krakow there was a separate session on the matter [6].
- Nomisma.org started as a small motivated team, however, it still took a long time and lots of argumentation to come to the point where we are right now. With its growing application areas, we are facing the problem that people still think in tables or spreadsheets. It requires time and effort to bring people on board and to help them to understand the extended possibilities (e.g. class system, multi inheritance, …) of ontologies. At the same time, we need to understand what demands and additional requirements they have. The communication between people is in fact the greatest challenge of all!