With the Shund.org corpus approaching 25,000 installments (from a single publication!), exploratory analysis has become a daunting prospect. Distant viewing and textual analysis techniques provide us with a way forward, but there are emerging alternatives to help us make sense of large and unwieldy data.
For example, virtual and augmented reality technology allows non-technical scholars the ability to physically navigate the digital corpora using digital representations of their head, arms, and body. Combined with head-mounted interfaces that support key visual cues associated with depth of field (e.g. stereoscopy, motion parallax, etc.), immersive visualization allows more efficient pattern, cluster, and detail recognition, among other benefits. Immersive tools like these afford intuitive, body-centered actions, like leaning in, or bending down, that you can’t do with a mouse or a screen.
We’ve been exploring this method using commercial data visualization software with virtual reality support, Virtualitics.As Virtualitics’ creator's state, “Massive and complex data sets – no matter how content-rich or how expensively obtained – are of no use if we cannot discover interesting patterns in them (Donalek et al. 2014)”. In Harvard’s Earth and Planetary Sciences’ Research and Teaching Visualization Laboratory, we used Virtualitics to study a 40-year (1900-1940) range of the Shund.org publication metadata, simultaneously mapping authorship, publication date, and – perhaps most interestingly – genre tags (e.g. novel, short story, sketch, etc.), the combination of which allows us to study the evolution of Shund in the early 20th century.
Indeed, Professor Zaritt was quite literally able to walk along a digital timeline representing these publication data, noting periods of author prolificacy and genre trends. Tools like Virtualitics represent a current generation of “productivity-grade” AR/VR software, allowing researchers the ability to import their own tabular data, prep that data for immersive visualization using a robust desktop client, and to engage with their data across a network of headsets (i.e. multiple people exploring the virtual field at once).
There are limitations of course. Individual data point representations are limited to simplistic shapes in Virtualitics, making the visualization of certain categorical features difficult. Importantly, that includes the full text contents of works in the database.