A New API for WCMA
Activating the Collection for faculty, students, artists and dreamers
It was a great summer at WCMA Digital. Summer means focused time to work on systems and infrastructure (yay!), without the delightful distraction of classes and students and faculty. It also means taking what you’ve learned in the hustle and bustle, and applying it, locking in gains for the future. For us, that means a new flexible datastore and a new API for our collection data.
First, some background on WCMA Digital: we’re a Mellon-funded initiative to inspire new use of the collection. Two essential questions power our work: How can we make WCMA’s collection more useful for students, faculty and our community? and How can those uses add context to the collection itself?
Our first step was to open our collection data. All of it. With a complete set of metadata available for download, we welcomed College collaborators in a new way. As word got out, collaborations led to projects, and projects taught us more about how we could make our data more useful. We posted a complete set of collection thumbnails to aid in visualization, computer vision and machine learning projects. We added object dimensions, standardized dates, and found opportunities to enhance and update our collection data.
Opening the collection opened a floodgate of interest, opportunities…and data. Stats classes want to use the collection data to study clustering algorithms! Great! Computer Science students are looking at keyword tagging with machine learning! Brilliant! That kind of use shows us what data is most useful. It also adds new context and knowledge to the collection. There’s no better way to improve your data than to have others use it.
We learned we needed a flexible datastore. We knew the core data we wanted to include, but we also knew we’d want to add new kinds of data next year (even next semester). It had to be easy to get data in, and easy to get data out. (Alas, museum collection management systems aren’t built for such.) We needed a system that would ingest data from many sources (and allow us to monitor and control the process). Give it a name and value, relate it to an object ID, and boom — it’s part of the data. That’s the idea.
The API was our next step. It’s a great thing to post collection data as a CSV on Github; for many digital humanities projects, it’s the best thing. But it also feels just a bit standoffish, a one-size-fits-all data buffet. Ideally we’d serve up fresh collection data in the moment, made-to-order in the most useful format. The buffet is available on Github for those who love it, but now we’ve got a fresh new GraphQL API for those who might enjoy a more full-service approach. (GraphQL improves the developer experience for us — you tell the server exactly what you need, and how you’d like to see it, potentially saving multiple API calls. And there’s a nifty API explorer for documentation.) The API is one way to make it as easy as we can for students, faculty and the community to use our collection data.
We were thrilled to partner with Micah Walter Studio to design and implement the datastore and API. (Thanks Micah! Elasticsearch and GraphQL FTW! I’ll leave it to Micah to share technical details in an upcoming post.) The team built on previous work developed with the Barnes Foundation for their online collection. (Thanks Shelley!)
Special thanks go to the WCMA Digital team as well. Collection Developer Jim Allison wrangled the TMS tables to get some beautiful JSON files into the datastore (article forthcoming), Associate Registrar Rachel Tassone guided the flow of collection data, and Liz Gallerani and Sonnet Coggins connected us to eager collaborators. The Mellon Foundation makes it all possible.
What’s next? We’re introducing the API to students and faculty, new friends at NEW INC., and you. We’re eager to see how it’s used in fall classes, projects and collaborations — and we’ll learn from seeing how it’s used. Our hope is not just that people will do computery things with our art collection, but also that people will do artful things with their computer applications. We’re building a web-based tool (atop our new API) that will help students and faculty visualize the collection from a range of perspectives. And we’re working on our third data-driven collection exhibition for 2019. Stay tuned!
