KEY LEARNING POINTS + PROFESSIONAL ISSUES
One of the key learning points in the process of creating a preliminary model for the MoMA data was navigating the various websites of the different ontologies. The Linked Open Vocabulary site was the primary resource for finding existing predicates that could be applicable to the model. Even if a predicate seemed to be applicable to a given relationship, it was necessary to visit the ontology website or, alternatively, view the downloadable RDF file linked to from the Linked Open Vocabulary predicate entry. The two main considerations when selecting predicates are ontology notability and constraints. Notability comes into play in that it is better to select a widely used predicate from a well-known ontology rather than an obscure predicate from an unknown ontology. Although it may be necessary to select a predicate from a lesser known ontology if it meets the needs of modeling a particular domain (the European Legislation Identifier (ELI) framework for modeling legislation, for example), better known ontologies are more likely to have strong documentation, be maintained, resulting in truly persistent URIs, and make your linked data more easily accessible and understandable by a wider audience.
While it is possible to go deep into the weeds of axioms and logical mechanisms of ontologies, the two main constraints to take into considerations when selecting predicates are domain and range. With the predicate acting as the relationship between the subject and an object in a linked data triple, the domain indicates which classes of subject a predicate can be used with, while the range identifies the allowable values or objects (What are Properties?, n.d.). While adhering to domain and range requirements had been more of an abstract consideration when thinking about linked data modeling prior to this fellowship, creating the conceptual model required me to dig a little deeper to understand which predicates could logically be used to relate two classes to one another. As a matter of documentation, I recorded the domain and range of each of the existing predicates selected for the preliminary model, which can be viewed on the Preliminary Model page of this report.
EVALUATION + CONCLUSION
This final report is the concluding deliverable for my time as the 2018-2019 Linked Open Data fellow at the Museum of Modern Art archives, under the supervision of Jonathan Lill, the Leon Levy Foundation Project Manager of the Museum of Modern Art Archives. It is my hope that the subsequent fellow in this position will refer to the work I have done for the purpose of refining, improving, or expanding upon the model to best suit MoMA's needs. This project is finished only insofar as my fellowship is bound by an "end date", but it is just the first iteration of modeling that can be improved and can then be used to implement the transformation of the relational database information into linked open data. To this effect, there is not a formal "evaluation" of the project with predetermined criteria or rubrics, although it will be iteratively evaluated, informally, as the project moves forward with the work done by the subsequent fellows.
A strength of this project has been the flexibility with which the scope and nuance of the model could be adjusted, in response to newly acquired knowledge or discussions with Jonathan Lill. An additional strength is that the "Exhibition Concept" and "Exhibition Instance" classes are grounded in the existing preeminent cultural heritage ontology (CIDOC CRM) and its extension in Linked Art, but the model in its entirety is not bound by these frameworks. A weakness of the project might be that the model does not encompass all of the classes of items that it eventually should (namely, art objects) and that there are likely other ontologies to be considered in relation to the model (for example, the Cultural Event Ontology). However, I do not deem these as insurmountable issues in the sense that a subsequent fellow could choose to investigate these avenues.
As an information and library science graduate student, my main objective has been to focus on knowledge organization. This fellowship provided the opportunity for me to hone my skills both in the sense of managing large amounts of data when retrieving and transforming art exhibition history data for the relational database, but also gave me the opportunity to dive head first into learning about how the CIDOC CRM is fashioned and how it is intended to function for the organization of information . This experience has been invaluable in providing a real-life need for becoming familiar with CIDOC CRM in order to make decisions about modeling a particular domain. While I do not believe that I will need to necessarily become an expert in CIDOC CRM, I am quite sure that it will come up again in my professional career as I seek to work in information management in the cultural heritage domain. It will be beneficial to already posses a familiarity with it in all of its complexities.