Friday, 2006-10-27, 11:00
Stéfan Sinclair, (Anthony Sapp), Stan Ruecker, “Experimental Reading: A Humanities Visualization Toolset for the TEI”
Ruecker began, then Sinclair demonstrated part of what they’d developed, which I found a nice reversal of the usual pattern. (Ruecker has the computing degree, Sinclair the literary one, though it’s clear that both are coders.)
Their focus is rich-prospect interface-browsing, arising in part from Ruecker’s work with the Orlando Project. Ruecker talked a bit about Appleton’s approach re: art history and James Gibson’s re: environmental psychology. Some points:
- There should be some amount of prospect (some representation) of each item in a collection.
- Offer tools for manipulating the display.
- The tools should rely on information emergent from the collection (e.g., encoding, or for Delegate Browser, participants’ photos).
- Where possible, > 1 form of representation should be available.
- The representations should link to additional data.
User study as a qualitative thing: Ruecker mentioned David Slees’s diagnostic–one should be able to do things with certain tools and data, and the results are iterated into the next design phase. Transferrability of design: equates to generalizable research in the sciences?
Three families of browsing interfaces: Watching the Script, Digital Playbook, and Mandala.
Watching the Script is a prototype of an online stage, which students, directors, etc. can use to learn lines, experiment with blocking, and so on. It uses .swf and offers a prospect of speaker involvement as a stripe down the left-hand side. (Good for seeing who can be cross-cast, e.g., since they’re not in a particular scene.)
Digital Playbook and Community Traffic are extensions of Watching the Script; they involve, respectively, the dynamic blocking of football (uh, soccer) team members and the simulation of traffic flow. Pretty effective demonstration of how different yet analogous constraints could be dealt with and rendered.
Mandala (here Sinclair took over) isn’t designed to be transferrable, really. It enables nuanced browsing of XML-encoded data via the definition of certain axes that “attract” data points visibly; it’s originally about collection structure. Phase I was built in Flash: all points are visible around the periphery of a circle, and selecting “speaker,” e.g., draws in things of that type; selecting something else instead frees the points that don’t match the second criterion (axis). [The visualization resembles very large iron filings interacting with magnets.] Flash doesn’t deal well with large collections, so phase II uses Java, with color-coded spindle lines. It enables the weighting of axes. Phase III integrates an earlier project Sinclair built called HyperPo, which allows real-time morphological analysis. [Missed something here about the utility of embedding features into other applications.]
Q&A
someone: How would one test success of a modeling scenario? response: as a relative improvement, perhaps, over current methods. [This issue returns in Quin’s keynote.]
[These three are from handwritten notes, as is the final keynote; in between I managed to type my notes.]