Library Hacking with the DPLA: Megacollections, Visual Search, and the Ecology of Metadata

Last week, Matthew Battles and I joined a group of literary hackers, digital humanists, media designers, and veteran librarians at the DPLA (Digital Public Library of America) appfest on the sprawling top floor of Chattanooga’s brutalist-era public library. We wrangled with library metadata all day, seeking new modes of engagement with unwieldy media collections integrated from institutions across the country. The DPLA just released a new API and is preparing to meet your queries with troves of collections information and content. The appfest was a challenge to those in the library community as well as techno-savvy sympathizers to prototype the first DPLA killer-apps. The forty or so people in attendance came from near and far: there was a strong showing of library staff local to Chattanooga and developers, designers, and media scholars flew in from Boston, New York, Texas and beyond (Toronto).

A range of technical and humanist skills were brought to bear on application prototypes developed in small groups. The results included a system for scoring the similarity of media records (winner of the Chattanooga library trophy), an online reference-ticketing system for asking DPLA-related questions, a mobile app for authoring new collections records, and many more beyond memory. Matthew and I collaborated with a group including local staff as well as visitors. For eight hours, with only short breaks for BBQ and cupcakes, we sought to compile a visual collections map for the DPLA, registering the scope and range of integrated contributions. Jeremy Throne came up with the name: biblioGrapher. Matthew wove together an overarching narrative. For my part, I hacked together some Processing code to produce a hierarchical graph of the current DPLA collections using the treeMappa library (See image above: the map shows nested rectangles corresponding to the relative sizes of contributed records. The white “holes” represent query results). Other members of the group created a wireframe interface for making DPLA queries using the collections map as an interface and developed use cases that librarians might expect to see. It was only the result of a day’s work, but already it suggested a promising means of engaging with aggregate collections: using a hierarchical map to contextualize queries and prompt further, and more informed exploration.

Back in Cambridge, we are making plans to continue work in this direction. In our Library Observatory project, we plan to further investigate the opportunities in tree maps for understanding and navigating large, heterogenous media collections. Indeed, this approach is promising not just for integrated library data, but for aggregated records from museums, arboreta, and a range of culturally significant collections. We believe that such maps hold insights to understanding what is gained and lost by bringing together the work of distributed cultures of collecting into online repositories like the DPLA.