Semantic Timeline Tools for History and Criticism

New concepts in the visualization of time-based
events are introduced and applied to the fields
of historiography and criticism. These techniques
(perpendicular timelines, dynamic confidence links, and time-slice relationship diagrams) extend the semantic power of timelines so that they can show the development of complex concepts and interpretations of underlying events. An interactive software tool called “TimeVis” illustrates these techniques with both 2D and 3D views.
History is a referent discipline. Later events build on earlier events, though in unpredictable and complicated
ways. Historiography and literary criticism are the
histories of accumulated comments on a subject. The
underlying history and literature (the “base events”)
occur in one era, and commentary and subsequent events
(“secondary events”) are added later. However,
commentary is not made in the same order as the base events; scholars might spend decades analyzing a
writer’s later works, and subsequently change emphasis to her earlier works.
Visualizing such referent-based relationships through
time is very difficult with a single, conventional
timeline. The concept of stacked timelines of different eras [Jen03] was introduced to align commentary and consequent events with their referents (Figure 1). This is useful when secondary events are evenly distributed, but less useful when they are concentrated on subsets of the base events. Crossing lines are difficult to interpret, and important early events can end up leading to a forest of arrows. What is more, the x-axes of the two timelines have no relation to each other. This lack of relation is in fact the cause of the criss-crossing lines.
This paper describes three new timeline techniques
that can be applied to the study of history, criticism, and other fields with a temporal or referent component. Each technique serves a different research need. Figure 1. Stacked semantic timeline shows
references to earlier timespans.
Techniques
“Perpendicular timelines” are like stacked timelines, but with the second turned 90 degrees, and a second dimension
added (Figure 2). The added dimension is the time
dimension of the first timeline. This means that secondary
events can be visually grouped by the base events that they refer to, yet also be ordered by their own time. In effect, perpendicular timelines allow each original event or topic to spin off its own timeline of commentary and follow-up, arranged perpendicular to the base events.
Figure 2. Perpendicular timeline shows subsequent development of older events.
“Dynamic confidence links” build on perpendicular
timelines, and provide interactive feedback. Each timeline within TimeVis can have one or more “time slices”, which are markers indicating the current point of interest in that timeline (Figure 3).
Figure 3. “Dynamic confidence links” are illuminated
as the focus in the secondary timeline is shifted (via
a slice marker), highlighting revelations of older events.
As we move the current time slice marker in the secondary
timeline, we are focusing on “what we knew at time t about the events in the base timeline”. As those
secondary events occurred in the real world, it changed our interpretation and understanding of the base events. (For example, when Boswell’s papers, thought to be lost, were discovered in the early 20th Century, they revealed details of his life that were formerly hidden.) These changed interpretations and understandings can be represented by gray event or concept markers in the base line, connected
to the revelatory events of the secondary timeline by gray
lines. When the current time slice marker passes a secondary event, its lines and the base events to which it connects can turn from gray to black, indicating that this was the time at which those facts or interpretations became more plausible. That is, the tool dynamically indicates our
confidence in different claims, as a function of time.
(Note that the inverse applies as well. If secondary events tend to reduce our confidence in earlier interpretations, those revelations can cause the base events to turn gray.)
Perpendicular timelines, and the dynamic confidence links they enable, can also be extended from two dimensions to three dimensions (Figure 4). Just as the move from one simple timeline to perpendicular timelines frees the secondary events to be organized both by topic and by time, the extension of perpendicular timelines into an as well as two other criteria.
Figure 4. 3D timeline, with two slice markers showing
time-slice relationship diagrams. This illustrates
the growing importance of the green actor
as we compare time T1 to time T2.
Both the base timeline (now a 3D timeline space) and the secondary timeline (also a 3D timeline space) have more flexibility. Rather than use strict definitions for the two free axes in the base 3D space, they can act as a 2D
surface for organizing the time-oriented topic/subject bars. By shuffling the time-oriented bars around, the links between them can become more understandable. This is similar to the behavior of force-directed 2D network
graph tools such as Visual Thesaurus.
The secondary 3D space can now be organized with one axis for its own events’ time, another axis for theme or subject, and the third axis for actor. This allows us to
illustrate concepts such as “who knew what when?”
“Time-slice relationship diagrams” show how different actors are related to each other at the time pointed to by the current time slice marker. If events are organized around the concept of actors, and laid out in a 3D space,
then a 2D time slice through the 3D space can show a relationship diagram, indicating the “small world”
connections from one actor to another. The diagram is
equivalent to looking down the time axis of the 3D space, backwards in time, with older connections appearing
distant (and thinner) and newer connections appearing large. As the current time slice marker is moved through
time, the user can see relationships forming and
decaying.
Applications
TimeVis is being used to visualize controversies and
cover-ups in history, including Watergate and the Dreyfus
Affair. It is being used to investigate the historiographical record of acceptance of Vertot’s Roman Revolutions (1719), and the literary response, over the centuries, to Boswell’s writings. Those studies should conclude shortly, and while no scholarly breakthroughs are to be expected, what should emerge is a set of visualizations (diagrams, videos, and data files) of use to students and researchers who seek to capture the big picture of such topics that stretch out over time.
References
Akaishi, M. & Okada, Y. (2004), “Time-tunnel: Visual Analysis Tool for Time-series Numerical Data and Its Aspects as Multimedia Presentation Tool”, Proc. of the Eighth International Conference on Information Visualization (IV’04).
Chittaro L., Combi C. (2001), “Representation of Temporal Intervals and Relations: Information
Visualization Aspects and their Evaluation”, Proc. of TIME-01: 8th International Symposium on
Temporal Representation and Reasoning, IEEE Press, Los Alamitos, CA, pp. 13-20.
Combi, C., Portoni, L., Pinciroli, F. (1999), “Visualizing
Temporal Clinical Data on the WWW”, Proc. Joint European Conference on Artificial Intelligence in
Medicine and Medical Decision Making (AIMDM’99), Springer, LNAI 1620, pp. 301-311.
Drucker, J., Nowviskie, B. (2003), “Temporal
Modelling”, ACH/ALLC 2003.
Jensen, M. (2003), “Visualizing Complex Semantic Timelines”, NewsBlip Technical Report 2003001, http://www.newsblip.com/tr/nbtr2003001.pdf .
Knight, B., Ma, J., and Nissan, E. (1998), “Representing
Temporal Knowledge In Legal Discourse, Law, Computers, and Artificial Intelligence”, Information and Communications Technology Law, Vol.7, No.3 (Special Issue on Formal Models Of Legal Time), pp.199-211.
Konchady, M. (1998), et al, “A Web based visualization
for documents”, Proc. of the 1998 Workshop on New Paradigms in Information Visualization and Manipulation, ACM PRess, New York, pp. 13-19.
Kullberg, R. (1995), “Dynamic Timelines: Visualizing Historical Information in Three Dimensions”, M.S. Thesis at the MIT Media Laboratory. http://robin.www.media.mit.edu/people/robin/thesis/ .
Kumar, V., Furuta, R. (1998), “Metadata Visualization for Digital Libraries: Interactive Timeline Editing
and Review”, Proceedings of the third ACM
Conference on Digital Libraries, p. 126-133.
Plaisant, C., Milash, B., Rose, A., Widoff, S.,
Shneiderman, B. (1996), “LifeLines: Visualizing
Personal Histories”, Proceedings of the CHI ’96 Conference on Human Factors in Computing
Systems, ACM Press, pp. 221-227.
Richter, H., Brotherton, J., Abowd, G. D., Truong, K. (1999), “A Multi-Scale Timeline Slider for Stream
Visualization and Control”, GVU Center, Georgia
Institute of Technology, Technical Report
GIT-GVU-99-30.
Robertson, B. (2000), “The Historical Event Markup and Linking Project”, http://www.heml.org.
Rockwell, G. (2003), “3D Timelines”, http://strange.mcmaster.ca/~grockwel/weblog/notes/000120.html.
Staley, D. (2002), “Computers, Visualization, and
History”, AHC Book Series, Sharpe, M.e.
Staley, D. (2000), “Historical Visualizations”, Journal of the Association of History and Computing, Vol. III, No. 3, http://mcel.pacificu.edu/JAHC/JAHCIII3/
P-REVIEWS/StaleyIII3.HTML .
Tufte, E.R. (2001), “The Visual Display of Quantitative Information”, Graphics Press, p.28.
Tominski, C., et al (2005), “3D Axes-Based Visualizations for Time Series Data”, Proc. of the Ninth International Conference on Information Visualization (IV’05).