ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD

A New Visualization Tool Designed as an Aid for Textual Exploration

Dominique Thiebaut

Department of Computer Science, Smith College, Northampton, MA, U.S.A.

Larry Owens

History Department, University of Massachusetts, Amherst, MA, U.S.A.

Keywords:

Visualization, Encyclopedia of computer science, Prefuse, Java, Network, Tree.

Abstract:

We present an original visualizer that allows users to travel through the network of pages of an early encyclo-

pedia of computer science. The purpose of this tool is to better understand the relationship between concepts of

computer science at an early stage in the development of the ﬁeld. The visualizer is written in Java, interfaces

to a database server, and sports two different graphical representations, a tree and a graph that are logically

connected. We present our design goals, our choices of implementations and the challenges encountered.

1 INTRODUCTION

How did the appearance of the Personal Computer

alter the world of computer science? How did the

different areas of research, the technology hot spots

change because of it? What areas of computer orga-

nization and architecture shifted importance because

of the entrance of the PC on the scene? What new

areas of interest and research resulted? Such ques-

tions are interesting for historians of science, and to

help better explore these fugitive and elusive ques-

tions we have designed VisNomad, a visualization

tool that displays graphically the pages of an ency-

clopedia of computer science. This tool is primar-

ily a thought tool (Sugiyama, 2002) for one to nav-

igate the network of the references linking the pages

of Ralston’s Encyclopedia of Computer Science. Two

different versions of the encyclopedia are used, one

published in 1976 (Ralston, 1976) and the other in

1983 (Ralston, 1983), respectively, and fortuitously

bracketing the birth of the PC revolution. The ability

of the human eye to discern patterns (Schneiderman,

1996), clusters, or hotspots in networks makes this

tool a valuable companion for historian explorers.

In this paper we concentrate on the visualization

aspects of this investigation. Interested readers should

see (Owens, 2009) for the report on the historical ex-

ploration.

The data visualized is the alphabetical listing of

entries of both editions of the encyclopedias. Each

entry of the encyclopedia forms an item in our data

set, and the references the entries make to other form

the links between the different entries. Each entry is

accompanied by the number of columns of text as-

sociated with it, the number of references it makes

to other entries, and the number of related subjects.

These quantities provide valuable information as the

full-text electronic version of the pages is not avail-

able.

VisNomad is written in Java, and uses Java Web-

start technology. It can be used as a stand-alone ap-

plication or as an applet accessible on the Web. The

data are stored in an SQL server, and are dynamically

retrieved by the visualizer as users roam the network

of encyclopedic entries.

The visualizer offers three different paths for ex-

ploring the entries of the encyclopedias:

• The ﬁrst one is a standard keyword search-box (`a

la Google),

• the second one is a three-level alphabetical list of

all the entries forming a tree, with each entry as

the root of a sub tree of entries that are directly

linked to it, and

• the third one is a two-dimensional graph depicting

entries connected by spring-loaded links ﬂoating

in a force-ﬁeld layout.

Figure 1 shows the full visualizer and the different

123

Thiebaut D. and Owens L. (2010).

ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD - A New Visualization Tool Designed as an Aid for Textual Exploration.

In Proceedings of the International Conference on Imaging Theory and Applications and International Conference on Information Visualization Theory

and Applications, pages 123-131

DOI: 10.5220/0002816201230131

 SciTePress

Figure 1: Visualizer window showing the entry for Interrupt, in a hierarchical tree on the left, the graph on the right, and the

search bar at the top.

components.

In the next subsection we present our objectives

for designing VisNomad. The subsection that follows

presents similar and related visualization projects and

highlights the differences with our approach.

1.1 Design Goals

Our design strategy is to create a system where the

user’s attention is taken logically top-down and from

left to right, in a simple but effective “1, 2, 3” ap-

proach where continuity of cognition is maintained.

We also want to follow Shneiderman’s long recom-

mended design guideline (Schneiderman, 1996) for

effective visualization: “Overview ﬁrst, zoom and ﬁl-

ter, then details-on-demand.”

The result should be a tool that sports Ware’s

stated objectives for effective information visualiza-

tion systems, which are reworded in our context be-

low.

• Comprehension: The visualization must provide

the ability to comprehend how the 400 entries in

the encyclopedia relate to each other, plus how

they aggregate, cluster, and contribute to form a

self-contained body of knowledge.

• Perception: The visualization must reveal proper-

ties of the relationships between computer science

concepts that were not anticipated.

• Quality control: The visualization must make it

easy to discover problems in the data. (This item

is of lower priority for us, as the data set is fairly

small and easy to validate.)

• Detail + Context: The visualization should facil-

itate understanding of small scale features in the

context of the large scale picture of the data.

• Interpretation: The visualization supports hy-

pothesis formation, leading to further investiga-

tion.

Finally, the aesthetics of the visualizer is an impor-

IVAPP 2010 - International Conference on Information Visualization Theory and Applications

124

tant component of its design (Sugiyama, 2002) and a

guiding force during its conception.

1.2 Data

The data are collected from Ralston’s encyclopedias

and consist of 476 entries linked by 2,333 references

for the 1976 edition (Ralston, 1976), and 551 entries

connected by 2,714 references for the second edition

(Ralston, 1983) of 1983. For each entry we also have

the number of columns of text it contains, the number

of references made to other entries, and its number of

related subjects.

1.3 Related Visualizations Projects

A good survey of the various techniques of visualiza-

tions for graphs can be found in (Herman et al., 2000),

and covers many of the types discussed here, includ-

ing force-ﬁeld layouts, radial graphs, and tree-maps.

A collection of documents connected to each other

by explicit links, such as references, form a Knowl-

edge Management System (Maier, 2007), also referred

to as KM system, or KMS. An encyclopedia is a

knowledge management system. According to Trier

(Trier, 2005), a KMS can include up to four different

types of knowledge entities: process/activities, docu-

ments, individuals, or topics. Our visualizer and its

data fall in the category of a topics KMS. KM focus-

ing on individuals and activities are refereed to as so-

cial networks, and many visualization tools have been

developed and proposed in this area in recent years.

Ontologies are also knowledge management systems,

close enough to our effort to warrant our interest in

visualization projects that target these systems.

Possibly the closest visualizing tool in spirits to

our own is Jellyﬁsh (Horn and Jenett, 2005). It is an

unpublishedproject that has been recognized and doc-

umented by several Web aggregation services, such

as Infovis-wiki.org, which specializes in cataloguing

interesting visualization tools. The Jellyﬁsh Website

states that “[t]he project should be seen as an experi-

ment, which deals with a dynamic interface. The pur-

pose [is] to remove a static, conventional design and

to achieve a playful interface.” Written in Processing

(Fry, 2008), the visualizer is open-source and allows

the user to interact with various “blobs,” or jellyﬁsh,

that ﬂoat on the screen and represent different areas

of the arts. Clicking on a jellyﬁsh transforms it into

a ring of nodes where each node represents an impor-

tant artist or work of art in a given ﬁeld. Each ﬁeld

is represented by a jellyﬁsh. The artists or art works

from a given ﬁeld, when selected by the user, may be-

come linked to other artists or art-works in other jel-

lyﬁsh, creating a growing network of connected blobs

on the screen. Additional information about graphic

elements is given in the form of pop-ups containing

short text descriptions. While playful, smooth, and

aesthetically pleasing, the visualizer can show only

a small number of jellyﬁsh on the screen. Further-

more the jellyﬁsh can contain only a few tens of nodes

and only explicitly selected relationships are visible,

making the geography of the full network not easily

understandable.

Several viewers presenting networks of words

have also been proposed and are of interest to us.

We select two representative examples, both designed

for the Wordnet project (Fellbaum, 1998): the Visual

Thesaurus, a commercial package, and Wordnet Ex-

plorer.

The Visual Thesaurus made by Thinkmap

(Thinkmap, 2007) is a visualizer that allows users to

specify a word in a search box, and in return dis-

plays a force-ﬁeld laid out graph. The graph shows

the word and its many derivations and deﬁnitions as

nodes connected by edges representing semantic con-

nections. The data mined by the visualizer are ex-

tracted from the Wordnet database (Fellbaum, 1998)

at Princeton. Users interact dynamically with the

graph, moving nodes around, and collapsing and ex-

panding the graph at will, in a very ﬂuid and intu-

itive way. Text boxes containing various deﬁnitions

pop up when words are clicked. Thinkmap has also

released several software development kits (SDK) to

help users create similar visualization for her own

data. Unfortunately both the visualizer and the SDK

collection are available only on a commercial basis.

Nevertheless, the Visual Thesaurus presents a pol-

ished interface and is a good example of an effective

word-based visualization worth emulating.

Also mining Princeton’s Wordnet database is

Wordnet Explorer which “applies visualization prin-

ciples to lexical semantics” (Collins, 2006). It is

richer in its offering of visualization types than its

many Wordnet-based counterparts, and allows users

to use spring-loaded, force-ﬁeld based layouts, ani-

mated radial graphs, and tree-maps to explore words

and semantic associations. This richness of the visu-

alizations reveals the difﬁculty of presenting in one

visualization the intricacies of the relationship exist-

ing in the data.

Skillmap (Meyer and Spiekermann, 2006) is a

KM system focusing on individuals, their skills, and

showing their organization in a hierarchical struc-

ture. People and skills form nodes of a graph, and

“who-knows-who” and “who-knows-what” relation-

ships form the edges between the nodes. The goal of

this tool is for managers to discover through visual

ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD - A New Visualization Tool Designed as an Aid for Textual

Exploration

125

means possible parallel developments in the same or-

ganization. The choice for displaying the network

of individuals is also a spring-loaded force-ﬁeld laid

out graph. The interface is smooth and playful, and

written in Java using the Prefuse tool kit (Heer et al.,

2005).

Possibly the most versatile, do-it-yourself visual-

ization system available today is IBM’s Many Eyes

(Viegas et al., 2007), which is a project aimed at creat-

ing out-of-the-box Web-based visualizations that can

be used by researchers to explore data sets in many

different visual forms, with the least amount of pro-

gramming. “The goal of [Many Eyes] is to support

collaboration around visualizations at a large scale

by fostering a social style of data analysis in which

visualizations not only serve as a discovery tool for

individuals but also as a medium to spur discussion

among users.” Many Eyes is a successful attempt at

making visualizing data a simple task for researchers,

and many variants of visualizations are available. Un-

fortunately, Many Eyes does not allow the integration

and linkage on one page of several different visualiza-

tions of the same data set, which is a feature we are

pursuing for our visualizer.

Finally, we mention word trees (Wattenberg and

Viegas, 2008) as an example of a hierarchical ex-

pression of a collection of words or sentences linked

together, visually close to our vertical tree represen-

tation of the links between the encyclopedia entries.

The algorithmics needed for Visnomad is simpler, as

the tree depth is ﬁxed for all entries, while that of

word trees is not, as the trees represent prose.

While all these different projects are rich in fea-

tures we seek, none offers a full set of the ones we

require: open-source code, smooth integration of dif-

ferent views of the data in a linked fashion, and Web

deployment.

We cover our implementation of VisNomad in the

next section, and continue with a section on valida-

tion. We conclude the paper with projected enhance-

ments for the visualizer.

2 THE GRAPHICAL USER

INTERFACE

VisNomad is written in Java and makes extensive

use of the Prefuse tool kit, which we have chosen

for its wide grass-root adoption in many visualiza-

tion projects, and the richness of data representations

it offers, including graphs, force-ﬁeld layout, ﬁsh-eye

views, trees, radial graphs and tree-maps. Also im-

portant for its adoption is its open-source distribution

policy, and the Java platform which makes it possible

to deploy as an applet on the Web, or as a stand-alone

application compatible with all three major comput-

ing platforms.

Our visualizer’s main window is divided into two

areas, as depicted in Figure 1, one vertical, on the

left-hand side, containing a tree, and one more wide

than tall, on the right-hand side, containing a two-

dimensional graph of the network of entries. The tree

is three-levels deep. Its root is a single point, repre-

senting the whole encyclopedia, and all the level-one

nodes, its children, form the alphabetical list of all the

entries in the encyclopedia. Each entry is in turn the

root of a sub-tree containing the alphabetical list of

the entries that either reference this root, or that are

referenced by this root, a concept we revisit in a later

paragraph.

Representing the encyclopedia as a vertical al-

phabetical list (tree) responds to a natural aspiration

for humans to understand and navigate a collection

of terms, and has been suggested by Sugiyama as

a structural rule (Sugiyama, 2002) that should be

adopted for data visualization of this type of infor-

mation.

The tree and the graph provide collaborative and

coordinated views for the user. The tree can be ex-

panded or reduced at will, while entries selected in

the tree can be made to appear in the right panel at the

center of a network showing all their direct neighbors,

as Figure 2 illustrates.

Our goal is to provide users with intuitive and log-

ical paths of exploration in a way not practically pos-

sible with the printed encyclopedia itself. One ap-

proach enables users to search for a particular key

word, reduce the tree to only the subtrees containing

that keyword, and explore any of the visible entries

in their network environment. Another approach al-

lows users to start exploring the encyclopedia by read-

ing the vertical alphabetically organized entries, re-

duce the tree to only the sub-tree of interest and then

display the graph equivalent to the sub-tree. These

two modes of exploration requires the user’s focus to

move from top to bottom, and then left to right, in a

logical and intuitive manner.

Smooth animation accompanies the collapse and

extension of the trees, as well as the organization of

the graph, including zooming in and out, preserving

continuity of cognition.

We now describe the different features of the vi-

sualizer, starting with the data encoding.

2.1 The Data

Each encyclopedia is stored in its own database in the

form of the adjacency matrix of the directed graph

IVAPP 2010 - International Conference on Information Visualization Theory and Applications

126

(a) Selecting “Interrupt” in the tree.

(b) Collapsed tree, and expansion of network of entries centered around “Scheduling Algorithm.”

Figure 2: Transitioning left to right: the user selects an entry in the tree, and expands it into a network.

created by all the entries as nodes, and all the refer-

ences between entries as edges.

2.2 Three-level Tree

The user can peruse the tree by adjusting a vertical

slider, and can select one particular sub-tree, rooted

at one of the entries. The leaves of the sub-tree con-

tain not only the entries referenced by the root of the

sub-tree, but also the entries referencing that root. We

have found through experimentation that discovery is

facilitated when we remove the direction from the ref-

erencing action. For example, if the Central Process-

ing Unit page references the Interrupt page, and if in

turns Interrupt references Deadlock, then both Cen-

tral Processing Unit and Deadlock will be listed as

ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD - A New Visualization Tool Designed as an Aid for Textual

Exploration

127

(a) Full Tree (b) Collapsed Tree

Figure 3: Left: the tree as ﬁrst presented to the user. Right: once the user clicks on one of the children of the interrupt entry.

children of Interrupt sub-tree.

When the user enters a keyword in the search bar,

all the subtrees that do not contain the keyword are

eliminated, leaving only subtrees that contain the key-

word, either in the root or leaf entries. All the actions

pertaining to operations intrinsic to the tree are per-

formed with the left mouse button.

The right mouse button is used to expand a node

of the tree into a spring-loaded graph layout in the

right panel of the window. Note that the above re-

mark about Interrupt, Central Processing Unit, and

Deadlock is true here as well, as the graph is directed.

The graph created is fully interactive, allowing users

to:

• zoom in and out of the graph,

• move the graph around,

• zoom the graph to ﬁt the window,

• move nodes around to untangle the graph,

• click on nodes to automatically add all their miss-

ing neighbors as new nodes, and

• backtrack through the history of node expansion

to return to a previous state of the graph.

Most of these features are leveraged and adapted

from the Prefuse tool-kit which provides many algo-

rithms for these graphic manipulations.

The visual identiﬁcation of community structures

in the data is effectively enhanced by the resulting

tool.

2.3 Visual Attributes

Several visual attributes adorn the nodes of the tree

and of the graph. For the tree nodes, we use two dif-

ferent colors for the border of each box surrounding

the entry and for the inside of the box. The border

color is used to identify nodes that may contain key-

words, or that have been visited by the user. The in-

side color is taken from a gradient of colors repre-

senting the betweenness centrality (Dekker, 2008) of

each node in the network, with darker colors corre-

sponding to more central nodes, and lighter colors for

peripheral nodes. The border color is used to iden-

tify the nodes that have been visited, creating a vi-

sual path of the exploration. Nodes also temporarily

change border color when they happen to be the node

currently under the mouse pointer, or any one of the

direct neighbors of that node. Highlighting the direct

IVAPP 2010 - International Conference on Information Visualization Theory and Applications

128

neighborhood of a node of interest is an effective way

of identifying neighborhoods.

Another visual attribute controls the size of each

node of the graph. Sugiyama’s semantic rule

(Sugiyama, 2002) sets the size of the nodes to reﬂect

the number of direct neighbors it has, that is the num-

ber of encyclopedia entries that reference it directly,

or the number of entries it references. With Visnomad

we use the out-degree of each node to control the size

of each node, indicating directly by its size whether

the entry lists many related-subjects in its text.

Finally we mention backtracking as a feature of

the visualizer that was requested by users. Backtrack-

ing allows users to “undo” their last node-expansion

operations in reverse order (LIFO) of trigger. A stack

is used to keep track of the state of the graph at vari-

ous points of the exploration.

2.4 Database Connectivity

Prefuse supports very simple database connectivity.

However, because of strict security issues, Java ap-

plications are extremely restricted in the way they ac-

cess remote resources, such as databases. Such access

requires the use and complex setup of certiﬁcates,

which we ﬁnd prohibitive for our purpose. Instead

we use a Php tunnel, i.e. a Web page in Php that acts

as a CGI script to obtain data from the database. The

Java engine passes Web access requests generated by

the visualizer to the CGI script, and receives back ta-

bles containing entries and links that are rendered as

subtrees or sub-graphs. Masquerading the access to a

remote database as a simple access to a Web URL is

effective and fast. Each of Ralston’s encyclopedias is

stored in its own database, the name of which is either

passed to the applet as a parameter embedded in the

host Web page, or selected at start up. Other param-

eters are passed to the visualizer from the host Web

page, including the color scheme used for rendering

the different data attributes.

In the next section we present an evaluation of our

visualizer.

3 EVALUATION

The constantly evolving and self-adjusting spring em-

bedded visual layout is pleasing to interact with, and

has been shown in published research to receive pos-

itive feedback from users (Meyer and Spiekermann,

2006) (Meyer et al., 2005).

One of the computing limitations is the time com-

plexity of force-based layouts. For a graph of N

nodes connected by E edges, the time complexity is

O(N lnN) or O(E), whichever is greater, making the

display of large graphs of more than a few hundred

nodes impractical and jittery, since animation is used

and many frames are displayed per second. Fortu-

nately, the distance ﬁltering feature we have imple-

mented allows the user to explore large graphs while

showing only a subset of it on the screen, and, over-

all, we have found that long exploration of the graph

remains a pleasant operation for the user.

3.1 Using the Visualizer as an Historical

Tool: Example and Perspective

In this section we report in the use of the tool by its

intended user: a historian. The data reported is sub-

jective, and limited mostly to one individual’s explo-

ration, as well as the presentation of the tool at profes-

sional meetings of the society of historians of science,

most recently (Owens, 2009).

Historians are fascinated by the origins of the per-

sonal computer. Given the fact that our understand-

ing of its origins is inﬂuenced by our current van-

tage point (In some sense we cheat in knowing the

end of the story!), we decided to use the ﬁrst two edi-

tions of Anthony Ralstons authoritative Encyclopedia

of Computer Science and Engineering to explore the

intellectual landscape of computer science at a time

when the PC was just emerging. The ﬁrst edition of

1976 appeared when virtually no PCs existed, the sec-

ond edition of 1983 two years after the appearance of

the IBM Personal Computer and the beginning of a

period of dramatic growth.

In principle, it seemed possible to acquire a sense

of the disciplinary “lay of the land” by reading both

editions cover to cover, taking note of the “Related

Subjects” listed at the head of each entry. Linkage

groups the near neighborhoods of especially inter-

esting entries were even sketched by hand in this ini-

tial exploration. It became apparent, however, that

exploration of any depth –to second- and third-order

neighborhoods and beyond– was practically impossi-

ble without digitization and the assistance of sophis-

ticated visualization techniques. The development of

VisNomad made it possible not only to visual these

further neighborhoods, but to “walk around” the en-

tire virtual space of these two encyclopedias in a

lively and convenient manner that reveals unantici-

pated perspectives and helps unravel the tangled web

of associations that deﬁned the intellectual landscape

of computing in the 70s and 80s.

Results of this initial reconnaissance have been

encouraging and surprising. In spite of its status as

one of the great inventions of the modern age, in the

company of Watts steam engine, Edisons incandes-

ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD - A New Visualization Tool Designed as an Aid for Textual

Exploration

129

Figure 4: A historian takes a walkabout from Interrupt to Input/Output Devices, traversing the universe of Computer Archi-

tecture.

cent lamp, and the transistor of Bardeen, Shockley,

and Brattain, the PC of recent accounts was nowhere

to be found. What we did ﬁnd in our promenades

is not the mythical Great Invention or its associated

Great Inventor, but a diffuse and growing fascina-

tion with machines that were becoming smaller, more

powerful, mobile, and user-friendly serving as in-

creasingly intelligent nodes in larger, more elaborate

networks.

This somewhat surprising observation, based on

the virtual exploration that VisNomad has made pos-

sible, suggests an alternative to the master narra-

tive that currently organizes much computing his-

tory. This is a machine-based account that starts

with Babbage and his Analytical Engine, proceeds

to Holleriths punched-card machinery and IBMs of-

ﬁce equipment, before passing on to the stars of

the story, World War Twos ENIAC and its machine

progeny EDVAC, Univac, and a growing family of

mainframes, minis, micros, and, eventually, the per-

sonal computer. The problem with this master nar-

rative is two-fold: ﬁrst, it ignores the fact that the

ideas and technologies relevant to personal comput-

ing were widely-diffused in the community; and, sec-

ond, it leaves time sharing, the Arpanet, Internet, and

World Wide Web awkwardly positioned on the mar-

gins of the central story.

An alternative account would take its cue from

James Benigers The Control Revolution (Beniger,

1986) and focus on the evolution of technological sys-

tems and networks, beginning with the telegraph, the

railroads, the emergence of universal power systems

like Chicagos Commonwealth Edison at the begin-

ning of the 20th century, AT&Ts national telephone

network, and the SAGE air defense system of the

early Cold War. Within this alternative narrative and

its ramifying networks–from WANs to LANs and the

computerization of ofﬁce systems, the awkwardness

of the master narrative disappears and the “PC” reap-

pears as a small, smart, and convenient node, a coher-

ent part of a larger story.

4 CONCLUSIONS

We present an original attempt at investigating his-

tory through visualization of the interrelationship that

exist between the entries of an encyclopedia. The ex-

ploration is supported by an interactive and dynamic

visualization written in Java. Users are presented with

two different means of perusing the entries, one hier-

IVAPP 2010 - International Conference on Information Visualization Theory and Applications

130

archical and alphabetically organized, the other two-

dimensional, interactive, and dynamic, presenting the

network of reference patterns existing between the

pages of the encyclopedia, and highlighting clusters,

neighborhoods, entries that are peripherals, or those

that have high degrees of connectivity, and that repre-

sent important subject matters.

Although subjective, our preliminary exploration

of the encyclopedias with the visualizer has shown

that it matches several of Ware’s objectives for effec-

tive visualization, in particular comprehension, per-

ception, and interpretation.

We leverage many features of the Prefuse tool-kit

and implement key recommended design rules that

have been proposed in the literature for effective, aes-

thetic and pleasant visualizations.

This tool is work in progress, and several enhance-

ments and options are planned as a response to early

user feedback. We list below some of the options con-

sidered.

• We implemented both Jordan and betweenness

centrality measures (Dekker, 2008) for each en-

try of the encyclopedia and selected the latter as

the one representing more intuitively the relative

importance of the various computer science con-

cepts present in the encyclopedia. A deeper anal-

ysis of other measures of centrality (such as max-

ﬂow, or closeness centralities (Sugiyama, 2002))

might prove better suited at representing the key

players and the shift of importance that different

editions of the encyclopedia might reveal.

• Obtain the licensing rights to the full-text of the

encyclopedias so that the contents of the entries

can be browsed in the visualizer, either inside a

separate window or panel, or as part of a pop-up

dialog.

• While Prefuse is a ﬁne tool-kit, it is signiﬁcantly

large, making the download of applets a time-

consuming activity. The development and main-

tenance of Prefuse has stopped, and a new Flash

version of the tool-kit has been released. Flash is

based on ActiveX, runs on a small footprint less

than a megabyte, supports vector graphics, and

many other features that make it integrate seam-

lessly in Web pages. Porting the visualizer to

Flash is a logical step for future development.

REFERENCES

Beniger, J. R. (1986). The Control Revolution: Technologi-

cal and Economic Origins of the Information Society.

Harvard University Press, Cambridge, MA.

Collins, C. (2006). Wordnet explorer: Apply-

ing visualization principles to lexical semantics.

http://www.cs.utoronto.ca/ ccollins/publications/.

Dekker, A. (2008). Centrality in social networks: Theo-

retical and simulation approaches. In SimTect, Mel-

bourne, Australia.

Fellbaum, C., editor (1998). WordNet: An Electronic Lexi-

cal Database. MIT Press, Cambridge, USA, 1st edi-

tion.

Fry, B. (2008). Visualizing Data. O’Reilly Media, Inc.

Heer, J., Card, S. K., and Landay, J. A. (2005). prefuse:

a toolkit for interactive information visualization. In

CHI 05, pages 421–430. ACM Press.

Herman, I., Society, I. C., Melanc¸on, G., and Marshall,

M. S. (2000). Graph visualization and navigation in

information visualization: a survey. IEEE Transac-

tions on Visualization and Computer Graphics, 6:24–

43.

Horn, C. and Jenett, F. (2005). Jellyﬁsh, and encyclopedia

of the arts. http://www.carohorn.de/jellyﬁsh.

Maier, R. (2007). Knowledge Management Systems: Infor-

mation And Communication Technologies for Knowl-

edge Management. Springer, Berlin, 3rd edition.

Meyer, B. and Spiekermann, S. (2006). Skillmap: dynamic

visualization of shared organizational context. pages

528–540.

Meyer, B., Spiekermann, S., and Hertlein, M. (2005).

Skillmap: identiﬁcation of unrecognized parallel de-

velopments and of communities of practice in dis-

tributed organizations. In First World Congress of

the International Federation for Systems Research

(IFSR2005).

Owens, L. (2009). Reports from computerville: Books, vir-

tual walkabouts, and professional identities. A paper

presented at the Special Workshop on Michael Ma-

honey and the History of Computing(s) at the SHOT

conference, Pittsburgh, PA.

Ralston, A., editor (1976). Encyclopedia of Computer Sci-

ence, 1st Ed. Petrocelli/Charter.

Ralston, A., editor (1983). Encyclopedia of Computer Sci-

ence, 2nd Ed. Van Nostrand Reinhold Co.

Schneiderman, B. (1996). The eyes have it: A task by data

type taxonomy for information visualization. Techni-

cal Report 13, UMD.

Sugiyama, K. (2002). Graph drawing and applications for

software and knowledge engineers. World Scientiﬁc,

Singapore.

Thinkmap (2007). The visual thesaurus.

http://www.visualthesaurus. com/.

Trier, M. (2005). A tool for it-supported visualization and

analysis of virtual communication networks in knowl-

edge communities. In Ferstl, O. K., Sinz, E. J., Eckert,

S., and Isselhorst, T., editors, Wirtschaftsinformatik,

pages 963–983. Heidelberg: Physica.

Viegas, F. B., Wattenberg, M., van Ham, F., Kriss, J., and

Mckeon, M. (2007). Manyeyes: a site for visualiza-

tion at internet scale. IEEE Trans Vis Comput Graph,

13(6):1121–1128.

Wattenberg, M. and Viegas, F. B. (2008). The word tree, an

interactive visual concordance.

ENCYCLOPEDIA WALKABOUTS WITH VISNOMAD - A New Visualization Tool Designed as an Aid for Textual

Exploration

131