SAH Blog


Digital research tools by Morgan Ng

Feb 18, 2011

I still distinctly remember a moment from my sophomore year of college when, peering through the university's library stacks, I was suddenly overcome by a sense of hopelessness. Each of the stacks' countless books, most now gathering dust, represented decades of scholarly labor. Reading them all would require many, many lifetimes to achieve. Even as historians continue to break into forgotten archives, leading technological innovators of our time have sought not only to add to this seemingly insurmountable glut of data, but also to equip us with tools for sorting through and making sense of it.

The December 2010 issue of JSAH addresses this issue as it pertains to our discipline. Mario Carpo and Kazys Varnelis summarize the uses of Google Books, Google Images and Flickr for scholarly research. These discussions reflect a larger trend dubbed "digital humanities." What are some of the possible future developments in this field?

Although proponents of digital humanities have long called for new ways of accessing texts, architectural historians also work with very different kinds of information: buildings, images, and objects. Google's text-based search functions, then, pose considerable limits to our object-based research. Nevertheless, the company has made efforts at refining the parameters of image searches. As Varnelis notes, searching "Villa Savoye" and setting the image type to "line drawing" yields not building photos but axonometric drawings, sketches and diagrams. Algorithms that detect images according to conventions of visual representation immediately turn up different--and potentially more useful--classes of images from those based on textual search terms alone.

What if we bypassed text-based searches altogether? Typing ‘Mona Lisa' into a search box yields images of the painting. But what if someone found a picture file of the painting and wanted to identify it? What if a search engine allowed someone to upload the image and instead search "backward" for the painting's title, artist and production date? Such software is already available, just not yet in wide use.

Flat objects like paintings are the easy part. Buildings, however, are dynamic spatial experiences, and every snapshot of these objects is different from another. Backward searches of this type are far more difficult. This difficulty extends to any object with multiple instantiations. If a biologist, for instance, uploads an image of a rare bird, can programmers develop software that identifies the species? (Think of the uses, for art historians with no training in botany, to identify painted flora!) Although, as Carpo points out, Google has eliminated text searches based on keywords (as found in conventional bibliographies), for now there seems no equivalent development for image databases. "Tagging" images by conventional or arbitrary categories remains the standard.

Finally, I'd like to suggest that we need better digital architectural databases. While SAHARA is a step toward collecting digital information on buildings, the number of available images (10,000 at the outset, culled from powerhouse institutions) has a long way to go before it becomes useful for scholars working on specialized topics. That many historians, including myself, are far more likely to look through Google, Flickr and Wikipedia suggests that SAHARA's "members only" policy is ultimately a liability.

This isn't just a utopian plea for democratic knowledge. It's a selfish proposal. In an age when every design student can whip up 3D models of buildings or GIS analyses in a matter of hours, SAHARA's collection is not just too small--it's outdated. How many students in their introductory digital drawing classes, I wonder, have built the same model of Villa Rotunda? And for each studio trip to Shanghai, Dubai and New York, the classes' careful photo documentation and digital site reconstructions become potentially powerful scholarly resources.

What if our own databases included such objects as well? What if historians in architecture departments collaborated regularly with digital media instructors, so that each year of incoming students could learn the required software while constructing digital models useful for our research? On the other hand, scholars would use their expertise to vet such objects and help with matters of identification, and so forth.

Almost fifteen years ago, Columbia University's art history department began a high-profile collaboration with the engineering school to launch public digital renderings of Amiens Cathedral, on the premise that digital modeling would offer fresh ways of analyzing a canonical structure. These are now everyday tools of the design student. If we pool join forces with architectural designers and enthusiasts--not just scholars--they might become equally commonplace for the historian.

 --Morgan Ng 

Comment

  1.