HistoGlobe
Teaching History Visually
Marcus Kossatz, Sebastian Utzig, Simon Schneegans, Felix Lauer, Tobias Westphal, Jens Geelhaar,
Bernd Froehlich and Patrick Riehmann
Bauhaus-Universit
¨
at Weimar, Bauhausstrasse 11, 99423 Weimar, Germany
Keywords:
Historical Visualization, Glyph-based Visualization, Time-dependent Visualization, Information Visualiza-
tion, Visualization Applications, Interface and Interaction Techniques for Visualization, GIS, HGIS.
Abstract:
HistoGlobe is an interactive historical geographic information system (HGIS) that provides students with
gathered and curated historical information for self-study and aids teachers during history classes. The sys-
tem visually integrates temporal and spatial aspects of historical events, as well as affiliations and alliances of
states, routes of historic people and groups, and finally, detailed multimedia information. HistoGlobe relies
on familiar interfaces such as globes and timelines but augments them with new techniques including directly
manipulable moving entities such as troops and a direction-preserving presentation of treaties and other col-
laborations based on routing lenses. A field study with 12th graders revealed an overall solid usability of the
system and inspired the development of further features.
1 INTRODUCTION
Traditional books and static maps about history
mostly rely on elaborate knowledge transfer in pro-
saic manner, but often they do not adequately sur-
vey long-term, dynamic and complex processes and
temporal changes visually. In contrast, HistoGlobe
is a historical geographic information system (HGIS)
combining the sciences of history the study of
when? – and geography – the study of where?. It em-
ploys modern web technologies to show historical de-
velopment in society, politics, and economics at given
times and periods in history (see Figure 1).
HistoGlobe is aimed at two specific target groups:
history teachers and students. It helps teachers to
present the content of their history lessons in an inter-
esting, informative, and interactive way and students
to better understand events and their context in his-
tory. Overall, our major contributions are:
An advanced map interface that is capable of vi-
sualizing changing border and region information
on either a 2D map or a 3D globe.
The hivent concept of visually incorporating ma-
jor historical happenings in time and space, in-
cluding detailed information and multimedia.
Advanced timeline concepts incorporating time
periods concerning historical topics and individ-
ual hivents.
Moving entities is a direct manipulative glyph-
based approach of providing route information
about individual people and groups.
Interactive and static routing lenses are an appeal-
ing approach to presenting treaties and other col-
laborations globally by preserving direction infor-
mation.
History teachers from a public school were involved
during the design and implementation process and, ul-
timately, invited us to conduct a field study on 12th
grade users during classtime, as well as during self-
study. Overall, most students were satisfied with
HistoGlobe because they found it easy and fun to use
and reported that it provides a good overview of his-
tory topics. The results encouraged us to proceed with
our approach, yet also revealed shortcomings and a
lack of some capabilities that were considered in the
further development process.
2 RELATED WORK
Best-known for visually embedding time-related in-
formation onto a static map is Charles Minard’s flow
map of Napoleon’s 1812 Russian Campaign (Minard,
1869) featured by Edvard Tufte (Tufte, 2001). Time
has only one dimension; however, it can be linear
Kossatz M., Utzig S., Schneegans S., Lauer F., Westphal T., Geelhaar J., Froehlich B. and Riehmann P.
HistoGlobe - Teaching History Visually.
DOI: 10.5220/0006103102010208
In Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2017), pages 201-208
ISBN: 978-989-758-228-8
Copyright
c
2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
201
Figure 1: HistoGlobe’s user interface currently focusing on Central Europe during the bipolar world era with its local and
particular manifestation of divided Germany (see below). The hivent list on the right outlines all historical events being crucial
during the Cold War such as the selected Limitation Talks (see also corresponding hivent box containing further information).
for a set of singular events or cyclical for recur-
ring events (Ott and Swiaczny, 2001). The preferred
technique for displaying linear time is the use of a
timeline; either in vertical (the World History Time-
line (Frank E. Smitha, n.d.)) or in horizontal direction
that allows dragging and zooming, e.g. The Life of
Nelson Mandela (Time Magazine, n.d.) created with
Timeline.js (Northwestern University Knight Lab,
n.d.). Google Earth can be used to show histori-
cal imagery (Google Inc., n.d.). Retrospectively, so-
cial indicators are often visualized using maps and
timelines, e.g. GapMinder World (Stiftelsen Gap-
minder, n.d.) and the World Bank Maps (World
Bank, n.d.). TimeMaps (TimeMaps Ltd., n.d.) shares
some basic concepts with HistoGlobe: at specific time
points the geopolitical state of the world is depicted
and certain historical events can be seen on both the
map and timeline. However, TimeMaps lacks link-
ing and interaction. A very interesting concept is the
space-time cube, which extends a regular planar map
into a third time-representing dimension (for exam-
ple Kraak (Kraak, 1988)). The idea of the contained
space path that connects two-dimensional location in-
formation to three-dimensional movement represen-
tation dates back to Lenntorp (Lenntorp, 1976).
A matter related to moving entities is direct ma-
nipulation of objects on screen (described later in
Section 5). Kimber (Kimber et al., 2007), Dragice-
vic (Dragicevic et al., 2008), and also Karrer (Kar-
rer et al., 2008) introduced a complementary space-
centric technique where an object trajectory is de-
tected during a video sequence and visualized as a
hint path on selection. Then the user can drag the ob-
ject along the trail constrained by its trajectory, which
sets the global time and thus moves all other ob-
jects along. Kimber’s approach, mainly dealing with
surveillance videos, additionally provides dragable
glyphs on a floor plan. DimpVis by Kondo (Kondo
and Collins, 2014) investigated this direct manipu-
lation approach in the field of information visualiza-
tion. While the pointer remains connected to the data,
smooth manipulation within spatiotemporal value do-
main can be made.
3 BASIC CONCEPTS
HistoGlobe is intended to convey historical knowl-
edge and geopolitical development in an informative,
fluent, and interactive manner. Several aspects of his-
torical data have to be addressed regarding this pur-
pose including date, place, involved parties, move-
ment profiles, alliances, and additional background
information. The user interface is carefully designed
by enhancing and mixing established linked methods
such as the map (or globe) interface for spatial dimen-
sions (latitude and longitude) with novel approaches,
such as a hivent (historical event) concept or the topic-
IVAPP 2017 - International Conference on Information Visualization Theory and Applications
202
Figure 2: Countries colored due to their membership in
NATO (blue) or Warsaw Pact (red).
extended timeline for orienting and navigating within
the temporal dimension. The map presents informa-
tion regarding a given date e.g. current region and
country borders, hivents that occurred at about that
time, moveable entities, etc. The countries’ color
mapping is primarily used to emphasize affiliation in-
formation (see Figure 2). Of course, not every sin-
gle country label can be shown at once, which is why
the countries are ranked size-wise and lower priority
labels that would overlap are hidden. Besides navi-
gating and zooming the map in regular mode, as well
as in full screen mode, a high contrast mode can be
toggled when working under problematic light condi-
tions in the classroom (see control buttons on the left
in Figure 1).
Timelines are a traditional yet appropriate way to
provide temporal orientation and navigation. We en-
hanced it to a topic bar. Temporal information is de-
picted in the lower part as an interactively zoomable
timeline (with levels of annual, multi-annual, decade,
and century intervals). The current date of the visu-
alization, the now marker, is marked in the center of
the timeline. Changing the current date by moving
the timeline or manually setting the date updates the
map by revealing territorial changes by animated tran-
sitions. Topic information is attached in an additional
upper layer to provide supplemental context about
global epochs, as well as longer lasting matters on a
local scale in the geographic regions currently visible.
We implemented a hierarchical approach where each
topic (e.g. “World War 2”) can have sub-topics (e.g.
“Blitzkrieg” or “Turning Point of the War”), as seen
in Figure 3. Some topics show additional features, for
example coloring of the countries based on their affil-
iation with a military alliance (as seen in Figure 2).
Providing a proper and legible labeling of the top-
ics turned out to be a surprisingly difficult problem
on a zoomable and moveable bar. Writing the topic
name in more than a single line would destroy the in-
tended layout since it reduces the space that should
ultimately remain on the map. Dynamically resizing
Figure 3: The upper row shows consecutive epochs of local
German history. The lower row consists of global eras that
provide their hierarchical structure in between. Below the
topics the zoomable timeline is attached.
the topic areas in a horizontal direction would result
in historically incorrect information due to visually
wrong ending times or overlapping topics. Leaving
out vowels or the middle of words and replacing the
letters with “... made the labels barely readable. The
most legible solution that we have found so far is man-
ually creating specific abbreviations for discretized
time zoom levels and replacing the end of the word
with “...”, if necessary, for the current zoom factor.
Linking elements of the topic bar and the maps are
the so-called hivents representing a spatio-temporal
relationship on a single occasion as historically sig-
nificant incidents at a certain point in time and space.
They are visualized at the location of occurrence on
the map, as well as on the timeline, using a small blue
glyph. Hivents occurring close to each other on the
map are grouped into a cluster marker that can expose
(on demand) the clustered hivents in a radial layout
and makes them selectable. Furthermore, they can be
filtered and searched for by for names, people, places,
and time in the upper right box that instantly presents
matching hivents. Below the search box, the regu-
lar hivent list depicts all hivents of the current topics
of the given time periods. Hivent list and search box
share the space on the right side, i.e. if the search box
is used, the hivent list folds down (not depicted).
Detailed information about the events are pre-
sented on demand in hivent boxes: the location, the
date, a short description about the event, and an af-
filiating background image. Additionally and option-
ally a link to the related Wikipedia article, as well as
directly-playable multimedia content, is available.
4 PILOT STUDY
A pilot test should give an impression whether
HistoGlobe is suitable as a teaching aid in the class-
room while imparting the historical context of a
hivent in our example, the content and the conse-
quences of four major conferences at the end of World
War II: Casablanca and Tehran 1943, Yalta and Pots-
dam 1945. The lecture was video recorded for obser-
vation and a semi-structural interview with the teacher
was conducted afterwards to gain more insight. Dur-
ing the lecture the teacher alternated between his own
HistoGlobe - Teaching History Visually
203
prepared teaching slides and HistoGlobe to explain
the context of the conferences by using the map. Al-
though the teacher needed only a limited subset of the
prototype’s features, he stated later that there was a
big advantage to having a navigable and zoomable
map for visualizing the territorial development of
countries and membership in alliances for other his-
torical topics.
For a second test we tried to gather impressions
about the usefulness of HistoGlobe as a self-education
tool for students. In a computer pool the students
were instructed to write an overview about the bipo-
lar world between 1945 and 1993 with a focus on the
cold and hot phases of the time. We observed the stu-
dents during the test and interviewed them later on,
asked about advantages and disadvantages, collected
suggestions for improvement of the visualization, and
again had the students complete a questionnaire. Sev-
eral interesting aspects were revealed. No questions
were asked by the students regarding the usage of
the software everyone was able to use it from the
beginning. The hivent list was the main navigation
element primarily used by all students. It was con-
sidered to be very helpful to know which event hap-
pened where and when in order to get detailed infor-
mation. Hence, the usage of the map was surprisingly
more passive, mostly driven by selecting the hivents
in the list. The information on the map was widely
understood, and the colored alliances were consid-
ered especially helpful. One requested feature was
to introduce cities, rivers, etc. on the map for bet-
ter orientation. The timeline and the topics bar were
used less, maybe since the “Bipolar World” topic only
spans four decades.
Figure 4: The usage of HistoGlobe as learning material for
the students.
Due to the exploratory task, the search function
was not used at all by the students. The students
did not have any specific keyword to search for. The
hivent boxes were highly appreciated as the main
source of information. Often students copied the con-
tent from the box directly into their notes (although
they were not allowed to). Interestingly, nearly half
of the students favored the high contrast mode even if
it was not really necessary with respect to the lighting
conditions and the monitors. Overall, the vividness of
the visualization and the chronology and visual im-
pressions of the hivents (especially in the list) were
considered by the students to be very helpful.
5 INTERACTING WITH MOVING
ENTITIES
As revealed by our evaluation, one important short-
coming of the early prototype was the lack of (as we
call it) moving entities. These include the where-
abouts or the habituations of a certain person during
his/her lifespan. It also means movements of groups
of persons in a shorter time period, such as troop
movements during armed conflicts, as well as massive
migration of entire tribes or nations. Minard’s (Mi-
nard, 1869) maps visualized time varying attributes
of a single entity (most prominently the size of the
troops) as width along its path.
Figure 5: Troops moved by (left) time bar manipulation or
(right) by dragging directly on map.
We aimed at visualizing multiple entities at once
by using a glyph-based approach in order to avoid
occlusion by widths of different paths. Thus, the
paths themselves have a constant width, whereas the
changes of attributes are addressed by modifying the
appearances of the circle-shaped glyphs representing
the troops. We introduced the idea of using “glyph
footsteps” at particular locations along the path. This
eases estimating and comparing glyph sizes while fol-
lowing a path.
In this manner the main challenge was aligning
glyphs on their paths and on the global time slider ac-
cordingly. Obviously, any manipulation on the time
bar must entail an immediate and automatic move-
ment of all glyphs along their respective paths. There-
fore the previously covered distance is depicted as a
IVAPP 2017 - International Conference on Information Visualization Theory and Applications
204
solid path line (see Figure 5a). We decided that, for
the sake of orientation and also for forecasting the fu-
ture parts of the path towards its destination (beyond
the current date one the time bar), the future segments
are drawn as dashed lines (see Figure 5b).
Controlling the position of the entity via the time-
bar while looking at a particular path was an indirect
task that resulted in unnecessary context switches.
We favored and implemented a direct manipula-
tion approach (inspired by Dragicevic (Dragicevic
et al., 2008), Karrer (Karrer et al., 2008) and also
Kimber (Kimber et al., 2007)), who introduced the
metaphor for controlling video playback directly by
grabbing objects depicted within the video frame) that
enables the user to drag the glyphs directly along
their paths which instantly affects the global date and
hence the positions of all other entities. Precise input
movement along the path is not necessary; the user is
guided by the path’s constraints.
Due to the lack of more precise routing infor-
mation, a path consists of a discrete minor number
of known locations that have to be interpolated tem-
porally and spatially in order to create a continuous
path and to determine an entity’s position for every
time step inbetween the dates of the known locations.
Two different possibilities for coping with this situa-
tion were tested: first, simply condoning interpolated,
yet inaccurate, positions when interaction is discon-
tinued (as implemented for the timebar). In contrast,
releasing the mouse when directly dragging a glyph
entails an automatic animation to its next valid posi-
tion, which again changes the global time and causes
the movement of other entities that once more might
end up in an inaccurate position, at which point we
stop the recurring calls and allow interpolated posi-
tions. Interrupting this cyclical behavior might ap-
pear inconsequential; however it prevents the system
(in the worst case) from moving all glyphs one after
another.
At best, this issue might be solved by gathering
more information about the entities and their move-
ments, which revealed to be barely possible since the
necessary information is rarely available in a reusable
form. Usually the information is unstructured, scat-
tered, and embedded in numerous text documents,
many of which cannot be processed with standard in-
formation retrieval tools.
6 ALLIANCE ROUTING LENSES
HistoGlobe already provided a possibility showing af-
filiations with alliances by coloring countries accord-
ingly. From suggestions during our field study we
noticed that this approach lacked in supporting bilat-
eral treaties. We implemented a solution for depict-
ing higher information density referring to historical
treaties and conflicts in combination with the three-
dimensional globe, which serves as advanced map
representation. “Correlates Of War” (Maoz, n.d.) of-
fers a wide variety of data concerning international
relations such as bilateral alliances among states and
countries between 1816 and 2012. Edges represent
bilateral treaties. Every active incoming edge in-
creases the radius of circular nodes accordingly so
the user can easily spot and distinguish very actively
linked countries and lesser ones.
The number of simultaneously active alliances
drawn onto the globe is tremendous (even for selected
epochs) and produces crossings, as well as occlusion.
Of course, every directly connected component is go-
ing to be highlighted if the pointer is hovered above
a particular node. This includes every outgoing edge
and its corresponding alliance partner node (see Fig-
ure 7 left). Additionally the user can turn off all non-
involved edges of the whole graph by clicking a node
or even selecting two alliance partners consecutively.
Although such filtering makes the graph less
dense, crossings between edges and non-associated
nodes still occur if an uninvolved third country
node lies underneath the direct path between two
alliance partners. In order to avoid such incor-
rect visual impressions, Holten (Holten, 2006) or
Buchin et al. (Buchin et al., 2011), for example, pro-
posed bundling techniques by using or generating
control points for spline-based paths which result in
a sort of branched structure that is considered to be
more visually appealing. Unfortunately, bending and
bundling edges eventually entails losing their direc-
tion information. Reliable predictions that limit the
area of possible target locations are only possible if
the direction towards a treaty partner remains straight.
Even if the target is out of sight while looking at an-
other region, we can count on our mental model in
such a familiar environment like the Earth’s globe.
Figure 6: Interactive lens-based edge distortion. (a) All
nodes out of scope. (b) Incident ignored edges on node in-
side scope.
HistoGlobe - Teaching History Visually
205
Therefore, we developed two techniques for
avoiding false implications by maintaining the basic
direction of an edge, globally, yet introduced a local
repelling mechanism in the proximity of third country
nodes. First, we evaluated an interactive lens-based
approach that distorts all edges around the mouse cur-
sor (see Figure 6a). The user can alter the repelling
strength and thereby change the lens’ scope. For hav-
ing the opportunity to focus on a particular node and
its alliances while moving the lens above it, neither
the node nor its adjacent edges are affected (see Fig-
ure 6 b). Using this property, one can easily distin-
guish true from false incidences when hovering over
a node.
An interactive approach is nice; however,
wouldn’t it be better to provide this capability for
all nodes at once? Advancing our approach auto-
matically generates a lens around every graph node
whereas the nodes’ radii are used to determine the
scope of the so-called local control points. Again, in-
cident edges associated with a node are not affected.
All other alliance connections circumvent each node
along the edge but follow their primary direction out
of the scope again (see Figure 7 right).
Besides a proper path construction itself, addi-
tional attributes had to be visualized: for instance, the
type of alliance (defense, neutrality, non-aggression,
entente, or a combination of them). Since less than six
combinations occurred in the dataset, we ultimately
used a simple additive color model based on respec-
tive base colors. Thus, one can easily spot the appear-
ance of enacted alliance types by observing whole
color regions on the map. However, we are aware
that this may not suffice for some future data sets con-
sisting of more types and combinations of alliances.
Furthermore, the user is provided with details about
making use of the hivent information windows.
Another important attribute that might be of inter-
est is the endurance of such a relationship between
two countries mapped onto the elevation of an edge
whose highest point is then at the center of a connec-
tion. In order to avoid edges elevated at the same level
following identical paths (if two states enact multi-
ple alliances during the same time period), arcs are
unfolded on both sides of the initial connection (see
Figure 8, top). The arcs can even be merged with our
lens-based approach; however, they are currently only
useful for certain viewing angles (see Figure 8, bot-
tom). Further development is going to improve the
concept by readjusting the lenses along their edges to
provide optimal view every time the point of view is
changed due to navigation.
7 DATA AND IMPLEMENTATION
DETAILS
The prototype is almost entirely implemented on
client side in Coffeescript/Javascript. For depicting
the map, the Leaflet library was integrated, whereas
all other components were developed from scratch.
All three-dimensional and globe-related interface el-
ements are realized by using Three.js, a web-based,
plug-in free, and a high level library which pro-
vides the use of WebGL. We organized perspectively-
distorted and screen-space based interface elements
into different rendered scenes. By casting rays from
the mouse pointer into these multiple scene layers, se-
lection and object manipulation is achieved. Shader
materials were employed to parallel process the con-
nections of the graph layer on the GPU.
Starting from Natural Earth Data (Natural Earth,
n.d.) we created and edited a novel dataset of his-
torical countries of most parts of Europe from 1871
until 1990. Available historical maps were first
geo-referenced onto present coordinates and then the
country borders were extracted using QGIS and stored
as individual JSON files. The labels are stored sepa-
rately due to independent changes of names and ge-
ometries. Each one is assigned to a priority that deter-
mines the label’s bounding box. Thus a label is only
shown if it does not collide with any bounding box.
HistoGlobe uses an event-based spatio-temporal
data model (Peuquet and Duan, 1995) for changing
country areas: for example, changing the member-
ship of a country (e.g. Alsace-Lorraine 1919 from the
German Empire to France) or for changing or intro-
ducing country borders following the splitting of two
countries (e.g. the border between Czech and Slovak
Republic after the dissolution of Czechoslovakia in
1991). Animated transitions lead the users’ attention.
Areas, labels, and transitions are preprocessed on the
server and loaded directly onto the client to avoid
client-server interaction during transitions. However,
if the date changes on timeline, the server is notified
to send all hivents that have occurred since the pre-
vious request. For each hivent the changes are exe-
cuted: on the map, transition areas and borders are
faded in and the related old and new areas and labels
are queued and processed by adding new areas and
labels while the old ones are faded from the map. In
order to prevent large amounts of changes on the map
if the timeline is moved far, a rule-out mechanism is
implemented. Areas and labels that would be added
in one change but deleted in another one are removed
from both lists because they would not contribute to
the current state. With this mechanism it is possible to
move the timeline at a high speed there and back and
IVAPP 2017 - International Conference on Information Visualization Theory and Applications
206
Figure 7: Highlighted alliance partners (left) without and (right) with location based repelling.
Figure 8: Arc groups (top). Combination of location based
repelling and arcs(bottom). View dependent.
always get a fast and consistent update on the map.
8 CONCLUSION AND FUTURE
WORK
HistoGlobe is intended to support teachers in explain-
ing complex historical matters, as well as guiding stu-
dents with curated content during self-study. There-
fore, the interactive web-based application integrates
familiar interfaces such as globes and timelines with
new visual concepts including hivents, directly ma-
nipulable moving entities, and direction-preserving
route lenses.
The conducted pilot study with 18 students led to
encouraging feedback, especially with respect to the
fluent spatio-temporal coordination of map and time-
line. The hivent concept, in particular, was greatly
appreciated since the students used it in our observa-
tion to speed up knowledge transfer by focusing on
the hivent list as kind of an overview (somehow by-
passing the timeline with its eras), a behavior that was
originally not intended.
Authoring capabilities have to be built in for
teachers to develop their own courses, customize ex-
isting ones, compile different lessons, or derive web-
based homework for the students. They should also
allow students to put together stories, record them,
and present them to others. However, even with
proper authoring tools, most information has to be
gathered and curated by people. We found that au-
tomatically extracting information suitable for our
needs is barely possible since the necessary informa-
tion is rarely available in a reusable form. Borders
of states or realms that frequently changed in the past
are traditionally depicted on historical maps that are
mostly not even digitally available. In this regard, re-
trieving information that characterized moving enti-
ties such as locations, directions, durations, and num-
ber of people involved is also very important. Usually
such information is unstructured, scattered, and em-
bedded in numerous text documents and requires the
development of custom information retrieval tools.
In conclusion, the current revision of HistoGlobe
HistoGlobe - Teaching History Visually
207
covers both the questions When? and Where? very
reasonably. An important issue to be solved in fu-
ture work addresses the Why? (did a hivent happen?)
question. This concerns both the automatic retrieval
of facts as described above and the design and im-
plementation of visualizations that could aid in clar-
ifying what hivents are utilized or affected by others
and to what extent. We believe that interactive visual
history teaching systems such as HistoGlobe can sig-
nificantly improve the knowledge transfer and the un-
derstanding of complex historical processes–besides
being much more engaging and fun than traditional
methods.
ACKNOWLEDGEMENTS
The authors would like to thank Sascha Dobschal,
Christian Dominka, Frej Bjon, Lene Ganschow, Fe-
lix Schmidt, Bastian Weber, Max Weber, Chris Hor-
nischer, Sebastian Kpsel, Andre Karge, and Dorit
Fauck for their contributions during the development
of HistoGlobe.
REFERENCES
Buchin, K., Speckmann, B., and Verbeek, K. (2011). Flow
map layout via spiral trees. Visualization and Com-
puter Graphics, IEEE Transactions on, 17(12):2536–
2544.
Dragicevic, P., Ramos, G., Bibliowitcz, J., Nowrouzezahrai,
D., Balakrishnan, R., and Singh, K. (2008). Video
browsing by direct manipulation. In Proceedings of
the SIGCHI Conference on Human Factors in Com-
puting Systems, CHI ’08, pages 237–246, New York,
NY, USA. ACM.
Frank E. Smitha (n.d.). World History Timeline.
http://www.fsmitha.com/t-index.html. [Online; ac-
cessed 2015-11-30].
Google Inc. (n.d.). Google Earth Historical Imagery.
http://www.google.com/earth/explore/showcase/
historical.html. [Online; accessed 2015-11-30].
Holten, D. (2006). Hierarchical edge bundles: Visualiza-
tion of adjacency relations in hierarchical data. IEEE
Transactions on Visualization and Computer Graph-
ics, 12(5):741–748.
Karrer, T., Weiss, M., Lee, E., and Borchers, J. (2008).
Dragon: A direct manipulation interface for frame-
accurate in-scene video navigation. In CHI ’08: Pro-
ceedings of the SIGCHI Conference on Human Fac-
tors in Computing Systems, pages 247–250, Florence,
Italy. ACM Press. Best CHI 2008 Note Award.
Kimber, D., Dunnigan, T., Girgensohn, A., Shipman, F.,
Turner, T., and Yang, T. (2007). Trailblazing: Video
playback control by direct object manipulation. In
Multimedia and Expo, 2007 IEEE International Con-
ference on, pages 1015–1018.
Kondo, B. and Collins, C. (2014). Dimpvis: Exploring
time-varying information visualizations by direct ma-
nipulation. IEEE Trans. on Visualization and Com-
puter Graphics (Proc. of the IEEE Conf. on Informa-
tion Visualization), 20(12).
Kraak, M. J. (1988). The space-time cube revisited from a
geovisualization perspective. Proceedings of the 21st
International Cartographic Conference, 1995.
Lenntorp, B. (1976). Paths in space-time environments: a
time-geographic study of movement possibilities of in-
dividuals. Meddelanden fr
˚
an Lunds universitets Ge-
ografiska institution: Avhandlingar. Royal University
of Lund, Dept. of Geography.
Maoz, Z. (n.d.). The Correlates of War Project.
http://www.correlatesofwar.org/. [Online; accessed
2015-11-30].
Minard, C. (1869). Carte figurative des pertes
successives en hommes de l’Armee Fran-
caise dans la campagne de Russie 1812-1813.
http://upload.wikimedia.org/wikipedia/commons/2/
29/Minard.png. [Online; accessed 2015-11-30].
Natural Earth (n.d.). 10m-cultural-vectors — Natural Earth.
http://www.naturalearthdata.com/downloads/10m-
cultural-vectors/. [Online; accessed 2015-11-30].
Northwestern University Knight Lab (n.d.). Timeline JS3
- Beautifully crafted timelines that are easy, and intu-
itive to use. https://timeline.knightlab.com/. [Online;
accessed 2015-11-30].
Ott, T. and Swiaczny, F. (2001). Time-Integrative Ge-
ographic Information Systems: Management and
Analysis of Spatio-Temporal Data. Number v. 1
in Time-integrative Geographic Information Systems:
Management and Analysis of Spatio-temporal Data.
Springer Berlin Heidelberg.
Peuquet, D. and Duan, N. (1995). An event-based spa-
tiotemporal data model (estdm) for temporal analysis
of geographical data. International Journal of Geo-
graphical Information Systems, 9(1):7–24.
Stiftelsen Gapminder (n.d.). GapMinder.
http://www.gapminder.org/world. [Online; accessed
2015-11-30].
Time Magazine (n.d.). Nelson Mandela’s Ex-
traordinary Life: An Interactive Timeline.
http://world.time.com/2013/12/05/nelson-mandelas-
extraordinary-life-an-interactive-timeline/. [Online;
accessed 2015-11-30].
TimeMaps Ltd. (n.d.). TimeMaps.
http://www.timemaps.com/history. [Online; ac-
cessed 2015-11-30].
Tufte, E. R. (2001). The Visual Display of Quantitative In-
formation. Graphics Press.
World Bank (n.d.). Life expectancy at birth, female (years).
http://data.worldbank.org/indicator/SP.DYN.LE00.
FE.IN?display=map. [Online; accessed 2015-11-30].
IVAPP 2017 - International Conference on Information Visualization Theory and Applications
208