Today, The Junto chats with David Riordan, Product Manager of NYPL Labs, about Building Inspector, a crowdsourced digital project that invites citizen cartographers to “help unlock New York City’s past by identifying buildings and other details on beautiful old maps.” Read on about the Vectorizor, how you can contribute to The New York City Space/Time Directory, and how NYPL is making the “Google Maps of the past.”
JUNTO: Digitization is one form of access, but it’s also great to see The New York Public Library fostering community engagement with archives through a “digital barn-raising” project like this one. What is Building Inspector, and how did you get the idea for it? Can you give us a sense of your digital workflow?
RIORDAN: Building Inspector is a game-like site that lets anyone help The New York Public Library turn our historical maps into historical data so we can build models of historical New York City that make it possible to access and use this information in ways that weren’t possible before. When a physical object gets digitized by NYPL, we take a high resolution image and create metadata for it, and like many places we have a central place of where they’re made available to the world. But for so many of these materials, such as our maps, they can be so much more than just pictures; we can restore some of what it represents—details like street names, or building numbers—in a digital form.
Our maps, once digitized, move automatically from our big institutional repository to the Map Warper (using the Digital Collections API, which is available to any member of the public), where NYPL staff and members of the public can let us know where on earth these maps are representing, and they can be imported into GIS (Geographic Information Systems) software. That makes it easier to compare different maps that all depict the same geographic area. From the Map Warper, we run select maps through a tool NYPL Labs created called the Vectorizer to generate the data we use for Building Inspector, and also generate Map Tiles, so the historical maps can work like web maps we use every day. And once they’ve been processed in Building Inspector and reviewed by our Map Division staff, we’ll import the data into our big searchable database of place, which for now we call The New York City Space/Time Directory. One day, hopefully in the not-too-distant future, The New York City Space/Time Directory will function like a Google Maps of the past, showing things like local restaurants, pictures, businesses and more for any given address.
JUNTO: How did you choose your software? And how do you orient participants who may not be “digital natives” to using it?
DR: With much of the software and tools we build, there are lots of existing, open source tools we build upon. For instance, Building Inspector takes advantage of the decades of work that have gone into the open source geospatial technology sector and uses tools that power web mapping applications that we all use every day, such as the PostGIS geospatial database, Leaflet (for displaying our maps), and TileStache (for rendering the map tiles). That said, we really try to focus on building projects and tools that feel well-designed, and can be used by non-technologists. Even digital natives get confused by poorly designed software. And so we try to start with a kernel of an interaction concept and move outward from there.
JUNTO: How do you create, review, and edit the crowd-sourced (meta)data?
DR: We’ve got a couple of approaches, but our latest crowdsourcing projects including Building Inspector borrow techniques from the citizen sciences, in particular the work of Zooniverse. They use an approach of showing the same thing to multiple people, and when enough people provide the same answer and a consensus is formed, it’s probably pretty good. In the case of Building Inspector, we show the same building to at least three people (after they’ve gone through the tutorial), and if more than 75% agree, we go with their agreed view. If there’s disagreement, we show it to more people, and when we’ve shown it to a significant number of people and can’t reach consensus, it goes onto NYPL staff for review.
JUNTO: For those starting out in the field of digital humanities, what kind of training do you recommend? When building interest or grant proposals, what models did you look to?
DR: We’ve been working closely with CUNY’s Graduate Center, which has an incredible digital humanities hothouse that they’re building. Their DH certificate program is really excellent, providing a really powerful survey of the field and some practical means folks can sink their teeth in. Right now the people who are standouts in the field are thoroughly skilled in technical and humanities disciplines, and while they may see the technology as the means to fully explore the humanities, they’re deeply grounded in both.
JUNTO: What’s next for Building Inspector and NYPL Labs? Is there a digital tool that you’d like to use, but it just doesn’t exist yet?
DR: What’s next for Building Inspector is making more maps available for inspection, and hopefully, more tasks. When Building Inspector first launched last fall, the only thing it tasked users with was checking the accuracy of a building footprint. A few months ago, we introduced a handful of new tasks, including the ability to fix building footprints, identify colors, and input street numbers. We hope to introduce more tasks soon. More broadly though, NYPL Labs will be working on projects that have to do with the entire life cycle of NYPL’s digitized materials, from digitization and metadata creation, to digital service and use.
As for digital tools I’d like to use, there are plenty of tools that have been built for entirely different fields that would be tremendous for humanities researchers, if we could make them available outside of their commercial contexts and could get our materials into a format that would work with them. Other things that hold a lot of promise right now are in the field of Deep Learning, where it seems like automatic image classification and handwriting recognition may be only a few years away. We have hundreds of hand-written letters in our collections, and transcribing all the text would take decades for people to do manually, so this technology would open up a whole host of new things we’ll be able to do at scale in the digital humanities.