SPARQL for Provenance Research

The remodeled GPI has made its provenance data more useful than ever through a shift from flat-file records to a linked data model, but fully realizing its potential requires some additional user orientation. The current Arches platform that the GPI relies on allows users to handle some of this added complexity through Advanced Search features, but some ways of filtering and manipulating data remain out of reach, at the moment. For example, a researcher might want to search across data models to find all artworks sold in Paris between 1820 and 1840 that involve a specific French dealer. In such cases, users can (with some education and practice) use the SPARQL query language to bridge the gap.

SPARQL and Arches are different environments—with different structures, philosophies, and purposes. To support SPARQL-based access, data is moved out of Arches and into a Linked Data environment—one similar to what Getty uses to provide Linked Open Data access to its Museum Collection.

So, what is Linked Open Data (LOD)? While we won’t provide a full introduction here, LOD is an approach to data encoding and storage that prioritizes common, agreed-upon data structures that promote sharing and connectivity. Typically, LOD centers around a “triple”(subject-object-predicate) data model, which allows for simple and complex connections to be made between separate bodies of data—sometimes within a single organization or across organizations and disciplines. Here are some external resources if you would like to familiarize yourself with LOD:

What is Linked Data?: This video is thirteen years old, but still gets the basic principles behind Linked Data right. A good first watch before moving on to other resources.
Introduction to the Principles of Linked Open Data: A good introduction to linked data from a humanities research perspective.
Using SPARQL to access Linked Open Data: A more in-depth exploration of the above. The downside is that one of the exemplar endpoints (the British Museum’s) has been deprecated.

Once the data exists and is made publicly available in an LOD format, a method for accessing it is required, along with a way of “querying” or searching the data. SPARQL (or, the “SPARQL Protocol and RDF Query Language”—yes, one acronym really is nested inside another) is a query language to help users navigate complicated, interlinked data resources. How does one person relate to an event? And how does that event relate to an object? What about a thousand persons, or a thousand objects? SPARQL can make those connections, and return results in formats like CSV tables. It can be a new, exciting, and integral part of a humanities researcher’s toolkit. This is not a formal course. Instead, it is a guided tour of selected SPARQL queries that are especially helpful for navigating the GPI’s interlinked data. By going through the queries in sequence, we hope you will gain a greater understanding of SPARQL.

[Lessons Under Construction]