Multilevel Displays and Document Blueprints: Dynamic Browsing Using XML Structures and Text Features

Abstract
In this paper, we discuss how researchers can benefit from
tools that allow them to work with visualizations that rely
on XML data at different levels of granularity. For these
visualizations, we propose using a single interface that draws
on the underlying structural information at both the collection
level and the level of the contents of the individual documents.
We compare two models of this kind of interactivity that are
the subjects of our current interface design and prototyping
activities. One of these systems is predicated on displays relying
on sequence, and the other on visualizing the structure of items
and facilitating their traversal.
Discussion
The common goal of these interface systems is to provide
researchers with experimental means of combining
overviews of a document with tools for manipulating the
display. First is our multilevel document visualization system
(Ruecker et al. 2005) that combines three or more simultaneous
displays, including a microtext column showing collection
items, a separate microtext column showing the contents of a
particular document, and a reading view of the document. These
different levels of display are combined with tools for selecting
and manipulating a portion of the text in a subsequent display.
The purpose of the multilevel system is to allow the user to
work with a digital document within the visual context of related
material. We have currently installed it as a feature in the
prototype for watching digital scripts, where the standard
reading view is supplemented with a digital stage that allows
the reader to see both the dynamic text playback and the
blocking of the characters as they move around the stage. We are currently carrying out a user study of actors using the system
to learn their lines.
In addition to the dynamic text, we have recently begun to
consider the possibilities of using the microtext columns as
opportunities for providing overviews that can provide further
data. For example, they might be colour-coded and re-organized
according to some useful principle. In the Watching the Script
demo, one form of colour-coding currently available – for plays
that have been encoded in XML that marks the character names
– is to help the reader differentiate between characters.
For purposes of a director planning the play, this overview
allows exploration of the various relationships of the characters
on the stage. It can be use to address questions such as who is
on stage when, and with who, and how many lines does each
actor have? By selectively applying colour only to the character
or characters of interest, and having the remainder display in
regular black text, the director has a tool for studying the entire
play from the perspective of the staging.
Another overview panel in the prototype provides a list of the
characters. As a simple display it does not provide many
affordances, but with the addition of some further information
and the ability to sort, this list also has potential benefits. For
example, if we give the user the opportunity of adding next to
each character the number of lines that character speaks, or a
total of time spent on stage, or both, and then rearranging the
characters according to those numbers, we have another tool
for use in understanding the play at an overview level, and for
planning production. Combined with the blocking tool, these
overviews and their related tools begin to create a tool suite
that we hope will prove useful to actors, directors, and students
of theatre.
The second prototype produces what we are calling a Document
Blueprint: a compact visual representation of the markup in a
document. This blueprint can be used to suggest encoding
particularities of a document by colour-coding tags and
attributes in flexible ways. The identity – and therefore
colour-coding – of sections can be defined in ways that ignore
certain nodes in an XML tree, including elements, attributes,
and text.
Our first application of this system is to generate a document
view where the table of contents is initially displayed in a
legible font size and the markup of sections are displayed in
small, blueprint mode (where visual cues from the text emerge
but the text itself is too small to be legible). The Document
Blueprint system allows the user to define which elements to
toggle between legible and blueprinting modes. Depending on
which elements the user chooses, the system provides a variety
of overviews that can help both in understanding and navigating
the document, much as the conventional table of contents has
always done, but with the important differences of dynamic
display and interactivity (Ruecker 2005).
Future Directions
For the multilevel display, we are currently working to
extend the affordances of the various overview items. It
may be useful, for instance, to allow the entire surface to be
temporarily subsumed by the overview that currently only
occupies one column. The more complex display would show
the data, still not at a reading view, but at a smaller level of
granularity. Colour-coding and reorganizing this form of display
could prove both interesting and useful.
The document blueprinting project is the youngest of our
prototype projects and will likely evolve the most in the coming
year. We are developing this system in collaboration with the
Orlando Project, which in its next phase will be producing
literary critical material in volume form. The dynamic table of
contents in this case will be used as a navigation aide for readers
interested in browsing through the extended prose of the three
volumes. As has been the case with each of the prototypes, we
will pursue an iterative cycle of design, prototyping, and
development, with user study and experiment at each stage of
the cycle.
Bibliography
Ruecker, Stan. "The Electronic Book Table of Contents as a
Research Tool." President’s Panel. Congress of the Humanities
and Social Sciences: Consortium for Computers in the
Humanities / Consortium pour Ordinateurs en Sciences
Humaines (COCH/COSH) Annual Conference. London,
Ontario. May 27- 30, 2005. 2005.
Ruecker, Stan, Eric Homich, and Sinclair, Stéfan. "Multi-level
Document Visualization." Visible Language 39.1 (2005).
Sinclair, Stéfan. "Computer-Assisted Reading: Reconceiving
Text Analysis." Literary & Linguistic Computing 18.2 (2003):
175-184. doi:10.1093/llc/18.2.175