THREE-DIMENSIONAL DATA VISUALIZATION OF
ELECTRONIC MILITARY INTELLIGENCE USING THE
PROJECT BROADSWORD SYSTEM
THESIS
Michael L. Goeringer, Lieutenant, USAF
AFIT/GCS/ENG/00M-08
DEPARTMENT OF THE AIR FORCE
AIR UNIVERSITY
AIR FORCE INSTITUTE OF TECHNOLOGY
Wright-Patterson Air Force Base, Ohio
APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.
20000815 197
AFIT/GCS/ENG/00M-08
THREE-DIMENSIONAL DATA VISUALIZATION OF ELECTRONIC
MILITARY INTELLIGENCE USING THE PROJECT BROADSWORD SYSTEM
THESIS
Presented to the Faculty
Department of Systems and Engineering Management
Graduate School of Engineering and Management
Air Force Institute of Technology
Air University
Air Education and Training Command
In Partial Fulfillment of the Requirements for the
Degree of Master of Science in Engineering and Environmental Management
Michael L. Goeringer, B.S.
Lieutenant, USAF
March 2000
APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.
The views expressed in this thesis are those of the author and do not reflect the
official policy or position of the Department of Defense or the U. S. Government.
AFIT/GCS/ENG/00M-08
THREE-DIMENSIONAL DATA VISUALIZATION OF
ELECTRONIC MILITARY INTELLIGENCE USING THE
PROJECT BROADSWORD SYSTEM
Michael L. Goeringer, B.S.,
Lieutenant, USAF
Approved:
’ a 7 SA'vr?
date
Dr. John J. Salerno
^cw OC
date
I l/Udhl'
Maj. I
8 tfA&r gooo
date
Acknowledgments
The research presented here was only possible through the trust, support and
faith of many individuals. First and foremost I would like to thank my wife Nicole
and our three children Sasha, Phillip and Tyler. Their understanding and support was
paramount in keeping my research focused and on track. The sacrifices they made
are beyond words. Thank you...I love you very much!
Special thanks goes to Dr. John Salerno and the entire Project Broadsword
team. By allowing me to conduct my research at the Air Force Research
Laboratory/Rome Research Site, Dr. Salerno provided me with an opportunity that
few students get the chance to even consider. I would also like to thank Mr. Michael
Wessing and his support staff for encouraging my work and handling the temporary
duty and travel related issues.
I would like to thank Lieutenant Colonel Timothy Jacobs for his
guidance. His trust and faith in my abilities is greatly appreciated. I would like to
thank Major Mike Talbert for not only having an interest in my work but also taking
the time to be part of my committee. Finally, I would like to thank Mr. David Buick
and Mr. Rodney Forbes for some serious diversions that helped me keep things in
perspective.
1
Table of Contents
Page
Acknowledgements . i
Table of Contents. ii
List of Figures. v
List of Tables. vi
List of Color Plates. vii
Abstract. viii
1. Introduction. 1
LI Overview. 1
1.2 Background. 3
1.3 Problem Statement. 3
1.4 Research Objectives. 4
1.5 Assumptions. 4
1.6 Scope and Limitations. 5
1.7 Research Summary. 5
1.8 Document Summary. 6
2. Intelligence and Data Visualization: Relevant Issues.7
2.1 Overview. 7
2.2 The Need for Intelligence. 7
2.3 The Project Broadsword Solution. 8
2.4 Data Visualization. 13
2.4.1 Color.15
2.4.2 Vision. 16
2.5 Conclusion.18
3. Methodology. 20
3.1 Data Exploration. 20
3.2 Software Development Process. 20
3.3 Visualization Considerations. 21
ii
3.4 Data Navigation. 22
3.5 Scenario Generation.22
4. Development Environment. 24
4.1 Environment.24
4.2 Target Environment. 24
4.3 Project Broadsword Code Base. 25
4.4 Model Approach. 25
4.5 Evaluation of Hardware Platforms. 26
4.6 Software Development Languages. 26
4.7 Automated Tools. 26
4.8 Graphics Rendering. 27
4.9 External Support. 27
4.10 Commercial Off the Shelf Products. 28
5. Design and Implementation. 30
5.1 Environment.30
5.2 Three-dimensional Visualization. 31
5.2.1 Three-dimensional Objects. 31
5.2.2 Multilevel Glyphs. 33
5.2.2.1 Level One Glyph.34
5.2.2.2 Level Two Glyph. 39
5.2.3 Level Of Detail. 40
5.3 Navigation. 44
5.3.1 Graphics Navigation. 45
5.3.2 Textural Navigation. 46
5.4 Multimedia. 50
5.5 VRML Plugin Verification. 50
5.6 Conclusion.50
6. Findings.51
6.1 Introduction.51
6.2 Experiment Setup.51
6.3 Experiment Results.52
6.3.1 Visual Presentation. 52
6.3.2 Visual Navigation. 55
6.3.3 Textual Data Navigation. 56
6.4 Conclusion.58
7. Conclusions and Recommendations. 59
7.1 Introduction.59
7.2 Conclusions.59
iii
7.2.1 Implementation Consideration.60
7.3 Further Research. 62
Color Plates. 65
Bibliography. 77
Vita. 78
IV
List of Figures
Figure Page
2.1 Textual Results.10
2.2 Timeline Results.11
2.3 Geospatial Map Results.12
2.4 Preattentive Processing.18
3.1 Experiment Scenario.23
5.1 Symbolic Glyphs.32
5.2 Multilevel Glyphs.34
5.3 Cylinder Shaped Level One Glyphs.35
5.4 Cube Shaped Level One Glyphs.36
5.5 Transparency as Designed.37
5.6 Transparency in Operational Environment.37
5.7 Redesigned Level One Glyph.39
5.8 Abstract Level Two Glyph.40
5.9 Model Based Level Two Glyph.41
5.10 Low Level of Detail Glyphs.41
5.11 Low Level Of Detail Glyphs on a VRML Landscape Map.42
5.12 Proximity Sensor Activation Rendering. 43
5.13 Products Represented by Gray Glyph.44
5.14 VRML Environment with Frames. 47
5.15 Pop-up Dialog Box. 48
5.16 Text Handling. 49
6.1 Glyph Representation.54
6.2 Two-dimensional Geospatial Map.54
6.3 Product Exploitation with BGONE. 57
7.1 Current Results Display. 61
7.2 BGONE Enhanced Results.62
v
List of Tables
Table Page
5.1 Intelligence Data Mapping.31
5.2 VRML Navigation.45
List of Color Plates
Plate Page
Plate 1: Geospatial Map Results. 65
Plate 2: Preattentive Processing.66
Plate 3: Symbolic Glyphs. 66
Plate 4: Multilevel Glyphs. 67
Plate 5: Cylinder Shaped Level One Glyphs. 67
Plate 6: Cube Shaped Level One Glyphs.68
Plate 7: Transparency as Designed. 68
Plate 8: Transparency in Operational Environment.69
Plate 9: Redesigned Level One Glyph.69
Plate 10: Abstract Level Two Glyph.70
Plate 11: Model Based Level Two Glyph.70
Plate 12: Low Level of Detail Glyphs.71
Plate 13: Low Level Of Detail Glyphs on a VRML Landscape Map.71
Plate 14: Proximity Sensor Activation Rendering.71
Plate 15: Products Represented by Gray Glyph.72
Plate 16: VRML Environment with Frames. 73
Plate 17: Text Handling. 74
Plate 18: Glyph Representation.75
Plate 19: Two-dimensional Geospatial Map.75
Plate 20: Current Results Display.76
Plate 21: BGONE Enhanced Results.76
vii
AFIT/GCS/ENG/00M-08
Abstract
Today’s military electronic infrastructure solves many problems while
creating others. Using computers, battlefield and global awareness is brought to bear
through the near real-time linking of sensor platforms from around the globe. These
intelligence networks produce vast amounts of data that must be parsed, interpreted,
digested and stored by information gathering systems. As the amount of intelligence
data continues to increase, these text-based systems become cumbersome and
inadequate. To ensure vital information is not overlooked or discovered too late,
other forms of intelligence product management and data navigation need to be
investigated.
This thesis explores procedures for enhancing the capabilities of Project
Broadsword, an intelligence data retrieval system, using three-dimensional data
visualization. Employing a stand-alone representative environment, this research
develops methods and techniques for overcoming problems encountered when
visualizing large quantities of data. Textual handling through graphical triggers is
also addressed.
The results of this research demonstrate the effects of utilizing three-
dimensional visual cues to organize, perceive, and navigate vast amounts of
intelligence products and their associated metadata. Conclusions drawn from this
research directly affect the next major release of the Project Broadsword system,
currently under development.
viii
THREE-DIMENSIONAL DATA VISUALIZATION OF ELECTRONIC
MILITARY INTELLIGENCE USING THE PROJECT BROADSWORD SYSTEM
1. Introduction
1.1 Overview
Information is power. This notation has been the crux of successful political
and military actions throughout recorded history. Measures of this power range from
the influence of shamans and priests who were well versed in the ways of the natural
world to General Schwartzkoff with his sophisticated network of surveillance
platforms during the Gulf War.
Today’s military decision-makers utilize data and information under the
concept of Command, Control, Communications, Computers and Intelligence (C 4 I).
The concept of “Command” encompasses all levels of authority from the
commander-in-chief to the airman in charge of a security detail. Each level requires
the key individual to receive and assess known situational information, build options,
and then execute his/her final decision. According to the Department of Defense’s
Dictionary of Military and Associated Terms . “Control” is defined as
That authority exercised by a commander over part of the activities of
subordinate organizations or other organizations not normally under his
command, which encompasses the responsibility for implementing orders or
directives. All or part of this authority may be transferred or delegated.
1
In this context, “Control” lends itself to the means by which information and
command functions are passed. “Communication” depicts the means by which full-
duplex information is moved. In essence this can be anything from runners and
carrier pigeons to fiber optic network lines. Today’s decision-makers no longer rely
on pulp media as their sole source of data and information. With the introduction of
the first computer by John V. Atanasoff in 1939, the need for and use of “Computers”
has grown rapidly. Through this electronic realm, we are able to collect, manage,
cross-reference, validate, meld, and examine intelligence even as the amount of data
rises exponentially. Finally, “Intelligence” is the collection, analysis, presentation,
and utilization of information on an area or subject of interest.
As identified by Lee Paschall, former Director, Defense Communications
Agency, the line between information needs and information overdose is a blurry one
that is difficult to define (1). Numerous Department of Defense programs and
research efforts have been put in place to try and calculate the needs of the decision¬
makers while understanding that human factors play a vital part in the equation.
This research examines the “I” in C 4 I, with respect to the needs, productivity,
and natural ability of the intelligence analyst, through interaction with an intelligence
data dissemination effort sponsored by the Air Force Research Laboratory (AFRL),
Rome Research Site. Known as Project Broadsword, this web technology-based
production level program is enhanced utilizing data visualization techniques.
Conclusions are drawn about whether the added visual cues increase the intelligence
analyst’s technical productivity while maintaining an easy to adapt and use computer
interface.
2
1.2 Background
In 1994 the Secretary of the Air Force, Dr. Sheila Widnall, and Chief of Staff,
General Ronald R. Fogleman, directed the Air Force Scientific Advisory Board
(SAB) to identify those technologies that will guarantee the air and space superiority
of the United States in the 21 st Century (2). To that end, a study was conducted in
1995 that identified those areas that are crucial for the United States to continue its
dominance of air and space. The final report emphasized the importance of a global
awareness capability for the warfighter. To meet this challenge head-on, Project
Broadsword was created at the AFRL, Rome Research Site Information Directorate in
New York State.
The main thrust of Project Broadsword is to allow the intelligence analyst
seamless access to vast amounts of heterogeneous data. Achieving this goal requires
varied layers of information dissemination and presentation through all levels of
command. To efficiently solve such real-world problems is one of the major
challenges facing the Air Force today. As stated in New World Vistas, “The
problems of the next decade are to identify the relevant databases, to devise methods
for collecting, analyzing, and correlating them, and to construct the needed
communication and distribution architectures.” (3)
1.3 Problem Statement
Since the advent of the digital computer in 1939 the amount of data and
information available to an intelligence analyst has grown exponentially. Digital
transfer of intelligence provides knowledge faster, in bulk, and allows for more
precise tactical decisions. Today’s military intelligence products, which can be
3
found in text, multimedia, and image form, are navigated and exploited through
textual interfaces. As the amount of intelligence data continues to increase, these
interfaces become cumbersome and inadequate. In order to ensure vital information
is not overlooked or discovered too late, other forms of product management and data
navigation need to be explored. Addressing the SAB’s Global Awareness agenda,
Project Broadsword has developed a universal textual interface for data source
management and has begun to explore the potential benefits of data visualization
using a two-dimensional approach.
1.4 Research Objectives
The primary purpose of this research effort is to create a representative
environment to examine ways by which Project Broadsword can offer management of
intelligence products through data visualization. Based on verbal request of those
attending the 1999 spring Project Broadsword user conference, this research is
directed towards visualization in the three-dimensional realm. Key areas of interest
include product representation utilizing three-dimensional glyphs as well as suitable
and effective navigational approaches. To fulfill this objective, three-dimensional
visual cues must offer the analyst an efficient and effective way of organizing,
understanding, and navigating vast amounts of intelligence products and their
associated metadata. The Project Broadsword verification and validation team holds
the principle role of determining the suitability and productiveness of this approach.
4
1.5 Assumptions
If the results of this thesis are to be integrated into future Project Broadsword
baselines, a level of backwards compatibility must be maintained to the version 3.0
release. Additionally, all work is constrained to the environment supplied within the
Project Broadsword development and integration vault. This is to fulfill functional
requirements and maintain primary security certification.
1.6 Scope and Limitations
This research effort explores an effective and efficient three-dimensional
visual representation of user requested products when utilizing the Project
Broadsword interface. This research narrows its scope to a limited user base and
provides a proof-of-concept environment only. No attempt has been made to produce
a production quality entity or provide tailored features to a particular Project
Broadsword user class.
1.7 Research Summary
This research explores data visualization techniques that enhance an analyst’s
exploitation capability when employing the Project Broadsword system. Using a
stand-alone representative environment, this research develops methods and
techniques for overcoming problems encountered when visualizing large quantities of
data. Textual handling through graphical triggers is also addressed. The results of
this research demonstrate the effects of utilizing three-dimensional visual cues to
organize, perceive, and navigate vast amounts of intelligence products and their
associated metadata.
5
1.8 Document Summary
The remainder of this document is divided into six chapters. Chapter 2
discusses relevant issues in intelligence and data visualization. This chapter also
explores the Project Broadsword solution to seamless intelligence exploitation from
dissimilar heterogeneous data sources. Chapter 3 presents the research process and
design methodology. Chapter 4 describes the research development environment.
Chapter 5 outlines design and implementation decisions made based on data
visualization, data navigation, and software development. Chapter 6 contains findings
and reviews the observations made during this research. Finally, Chapter 7
summarizes this research, identifying specific contributions and proposed
recommendations for future work.
6
2. Intelligence and Data Visualization: Relevant Issues
2.1 Overview
The exploitation of military intelligence using data visualization is a relatively
new technique that few implementations have developed and put to practical use.
This chapter begins by providing a cursory review of the need for military
intelligence. It then introduces Project Broadsword, a software-based data retrieval
system that is able to access multiple intelligence product domains through a single
user interface. Finally, elements of data visualization that directly effect this research
effort are presented.
2.2 The Need for Intelligence
The basic function of intelligence is to provide the requisite support for
timely, sound decisions of all matters both in and out of a military conflict. As stated
by William E. Colby, “The purpose of intelligence is to help you act so that you can
have a better rather than a worse future. And if you act intelligently, and cause a
change in that future, then of course the prediction turns out to be wrong—for the
right reason, and you’ve really capitalized on what intelligence is all about.” (4)
Intelligence is performed in a cycle that consists of collection, production,
and dissemination of information. With this in mind, the tasking laid before the
analysts is to understand and adapt to the operational environment. Information
about the environment must be collected, analyzed, and finally molded into a
comprehensible output for use by a decision-maker. The importance of this process
cannot be overstated. As emphasized by General Ronald R. Fogleman, the ability to
7
collect information, rapidly correlate it, and then quickly disseminate it to the users
and commanders changes the nature of warfare (1). As we approach the 21 st century,
we teeter on the ability to locate, track, and catalog anything on the face of the Earth
in near real time. This capability brings with it an ever-increasing demand on the part
of the analyst to understand and ingest more data at a faster rate. When dealing with
these vast amounts of data, we must control the flow such that only the important or
key elements are brought forth and presented. Thus, only what is immediately
important is driven to the attention of the analyst. Based on these concerns,
intelligence exploitation systems are continually exploring advanced data presentation
techniques to maintain equilibrium between our need to get more information faster
and our ability to handle it.
2.3 The Project Broadsword Solution
Project Broadsword is an Air Force Research Laboratory effort sponsored by
the Joint Intelligence Virtual Architecture (JIVA), the National Imagery and Mapping
Agency (NIMA) and the Air Staff. In performance of their duties, the intelligence
analyst has four types of information to collect and exploit - imagery, video, text, and
audio. The mission of Project Broadsword is to make retrieval of these product types
as seamless as possible. In order to accomplish this feat, a set of tools and services
are employed which allow an analyst to search and retrieve information from a
collection of heterogeneous data sources. These sources include, but are not limited
to, the Military Intelligence Database (MIDB), the Image Product Library/Image
Product Archive (IPL/IPA) and the Automated Message Handling System (AMHS).
8
Project Broadsword is made up of two elements: the client and the
Gatekeeper. The client provides universal access to all available intelligence data
sources via the Gatekeeper by way of standard Internet protocols. This approach
allows the analyst to use a common web browser for system access. The Gatekeeper,
a robust thin layer of software, performs a variety of internal functions, most notably
the ability to simultaneously query multiple dissimilar intelligence databases while
returning uniform results. These results are available in a multitude of formats, to
include text listings, a chronological timeline, and a two-dimensional visual display.
Results are presented in one of three ways. First, all results may be returned
as textual descriptions as seen in Figure 2.1, with accompanying thumbnail images
(where applicable) similar to today’s modem commercial Internet search engines.
Figure 2.1 shows a product with a thumbnail image and the data elements under
which it was cataloged. These data elements, known as metadata, consist of
descriptions that, either individually or when combined, uniquely identify a product.
Figure 2.1 displays only a small sample of the possible metadata elements available
for a product.
If textual descriptions are not desired, a timeline may be chosen, as seen in
Figure 2.2, which displays all products in chronological order of creation. This
method organizes products by date and allows a larger number of products to be
displayed in the same amount of screen space. The results are limited to thumbnail
images coupled with a hypertext link to their associated metadata. Selecting a
hypertext link generates a window containing the selected product metadata, similar
to Figure 2.1. Figure 2.3 shows the third presentation method, a two-dimensional
9
SEARCH TOOLS
QUERY:
PRD.PRODFMT =
"TEFF6.0"
Revise Query
Save Query As
IPLtest2
[I PL 1.0 at Sun via Neptune]
Document ID IPA^sun_3^ 6365izMay97_725033
Security Classification Unclassified
Agency root
Publication Date 19970530123554
Format Image: TIFF 6.0 (.tifl)
Product Size 263,004 Bytes
Product Short Name ANIMG
Title IPLtest2
Post Date 19970530163659
Number Of Thumbnails 1
■ j r. r _ - . — '.• , ■ • ._ij.iCL,... v v*
siwiiiii ismmMmm
Compression Generation 0
Image Compression Not Compressed
Image Number 1
Thumbnail Image
Figure 2.1: Textual Results
10
SEARCH TOOLS
QUERY:
Revise Query
y ujdivi . _
PRD.PRODFMT =
"T1EE6.0" Save Query As
Source _I_ Query Status
TOTAL HITS 10
10 Hits
1997
| 1998 [
1999
n
[IPL 1.0 at Sun via Neptune]
r (U)cattest_3 311
[IPL 1.0 at Sun via Neptune]
HH -L
1 . L;-.-' -v -* v-.
p’ ‘-v.
aggMMM
■“■“^DisDlav Details
Displav Details
r (U)IPLtest2
r (U)cattest 3316
[IPL 1.0 at Sun yia:Neptune]
[IPL 1.0 at Sun via Neptune]
119970530123400 - 19990331194859
Enter Desired Time Window
Begin:
YYYYMMDDhhmmss
End:
|19970530123400
Zoom: | AUTO "3 [Current: Years]
YYYYMMDDhhmmss
J19 9 9 0 3 3119 4 8 5 9
Update Display |
ypertext Link
Timeline Heading,
Figure 2.2: Timeline Results
11
QUERY:
PRD.PRODFMT
"TIFF6.0"
Revise Query
Save Query As
Green Two-dimensional Icon
Figure 2.3: Geospatial Map Results (Color Plate 1)
12
geospatial map that graphically depicts the geographic location of each product by
use of a green square icon. This icon represents all products that were found at a
single location. The three icons in Figure 2.3 represent at least 3 products. They may
also represent three hundred products or more.
To take full advantage of Project Broadsword’s potential, data visualization
techniques must be further refined and exploited. The current two-dimensional map
offers a data visualization capability using geospatial coordinates as its base. As can
be seen in Figure 2.3, this does not provide any meaningful information beyond data
clustering. Since a single icon can represent an unlimited number of products, each is
a hypertext link, that when selected, results in a textual output as seen in Figure 2.1.
Being forced to return to a textual format immediately after results are displayed
negates the true essence of utilizing data visualization in the first place.
2.4 Data Visualization
The increasing number of data sources and information repositories coupled
with the availability of fast digital network access through the Internet and more
secure global networks has created an enormous demand for querying, accessing,
retrieving and understanding of information. One approach to aid in the organization
of these tasks involves data visualization. The effective use of visualization
addresses the challenge of discovering and exploiting information that may have
otherwise been overlooked or neglected.
Data visualization enhances decision-making through spatial exploitation of
diverse data products. Technological advances in computing have widened the data
visualization spectrum to a point where the greatest challenge is no longer the
13
available technology but rather retaining the intimate involvement and understanding
of the end-user. Traditional text interfaces, which navigate with a mouse or
keyboard, allow us to work on a computer, interacting with the system and not the
data. Data visualization, on the other hand, enables us to work with the computer by
enabling interaction with the data. This approach allows the system to be an
extension of our visual and cognitive system, creating a seemingly free-flow
convergence of user and data.
The quest for the perfect data visualization implementation is a moving target
and presents some major challenges. The first task is to locate and retrieve the
relevant data. In our world of dynamically changing and diverse heterogeneous data
sources, this in itself is a major undertaking. Though not generally a visualization
function, data retrieval is a required step toward a successful visual product.
Next is the challenge of understanding the data at a level that is appropriate
and comfortable for the user. Consider the different levels of military command, each
having their own intelligence requirements, yet accessing the same data sources.
Identical data elements may have a varying degree of importance at each level and
therefore require a different priority in the visual realm. Maintaining a high level of
human involvement and participation in the data processing and analysis
requirements through visual interaction supports this association. The intimate
involvement and understanding of data analysis significantly aids in the prevention of
erroneous abstract conclusions by ensuring the meaning and intent of the data is
portrayed in context of the task at hand, and not simply interpreted blindly. For
example, the sudden grounding of several Minuteman III Payload Transporters (used
14
to transport missile guidance components) may seem suspicious by an analyst who
does not understand Intercontinental Ballistic Missiles (ICBM) inspection
requirements. Effectively communicating the proper meaning of the information is
paramount. Portraying a perfectly crafted visualization is meaningless if the intended
audience does not understand or misinterprets it.
A further challenge of information visualization is the available hardware
interfaces and software display formats. The goal is to maximize the information
content within our physical constraints while accommodating and capitalizing on the
natural capability of the user. From a hardware perspective this is simply an issue of
resolution and rendering performance. For the user, complications such as visual
acuity and level of understanding must be considered.
2.4.1 Color
Based on research performed by the Department of Computer Science at the
University of British Columbia, an important factor to consider when building visual
models is the correct use of color (5). Color is a popular and effective attribute used
to map a dataset into a visual plane. Previous studies attest to the fact that choosing a
correct color, or number of colors, is an important step towards a successful
visualization. Research on the design of military systems suggests that the general
guideline for computer-generated images is no more than five to seven colors at a
time (5). Additionally, recent work at the IBM Thomas J. Watson Research Center
has focused on a rule-based visualization tool that considers how a user perceives
visual features such as hue, luminance, height, and so forth (6, 7). This work
emphasizes the understanding of data visualization; however, it has not addressed
15
how color aids in a more accurate identification of individual data elements.
The use of color is a powerful medium element that capitalizes on each
individual’s hard-wired perceptual capability. Color perception is deeply wired into
the human psyche and as a result our perception of color is fast, accurate, automatic,
and effortless. A well-constructed image will use color to reduce the amount of
thought process required to understand the information being presented. This thought
process, defines those elements required for more complex recognition, such as
reading text or attaching meaning to an object. Thinking is a relatively slow process
that requires mental resources and is error-prone. Misreading a word or sentence is
relatively common whereas mistaking a red apple for blue is not. Even those who are
colorblind easily acknowledge subtle changes in color. Based on these founding
principles, color tends to be the primary focus of data visualizations.
In three-dimensional space, color is an effective tool for the control of spatial
perception (8). In a natural scene, mountains off in the distance usually appear blue-
violet and indistinct. Known as aerial perspective, this visual effect is produced by
white sunlight hitting the atmosphere and refracting on its shorter wavelength (8).
Since the eye automatically interprets blue-violet and loss of detail as a sign of
distance, it can be creatively used to add a feeling of depth to a flat display. To
emphasize closeness, warm colors such as red, yellow, and orange are used.
2.4.2 Vision
Effective graphics development requires a general understanding of vision and
the distinction between “attentive” and “preattentive” processing (8). At the instance
16
a scene is first viewed, there is more information presented than can be cognitively
processed. To overcome this, we instinctively decompose the image into more
manageable pieces and then examine each piece individually. This divide and
conquer strategy is called attentive processing and is accomplished by moving the
eyes such that the current area of interest forms an image on the fovea, the highest
acuity area of the retina. Attentive processing can be thought of as a spotlight, which
the viewer moves across the scene. Once formed, the highlighted segments are
processed for further meaning, such as understanding an object’s representation, and
are no longer simply patches of color. In this attentive processing stage, the viewer is
still dimly aware of the sensations outside of the spotlighted area but has no idea of
their meaning. While focusing on the local area the viewer still maintains a sense of
awareness of the rest of the world.
To effectively perform attentive processing the user must effectively break the
image into useful segments and then decide which area to examine first. Preattentive
processing provides this function by relying on raw sensory information, such as
color, which has no immediate inherent meaning. As the user organizes and links
different colors encountered, natural image segmentation takes place. This
instinctiveness allows a similar color in geographically separate areas to reinforce
commonality. Figure 2.4 illustrates how a designer can use color to push a user’s
view in a desired direction (8). Notice how the red squares on the right immediately
draw attention from the dark blue squares and involuntarily force you to link them
into a single line. The lighter blue squares on the left differ only in brightness and
therefore the effect is not as strong. Both of these examples demonstrate how a clever
17
Figure 2.4: Preattentive Processing (Color Plate 2)
designer can use preattentive processing to draw a user’s eye to a desired location,
and further, control to some extent, the movement of the eye around an image.
The importance of optimizing preattentive processing and minimizing
thinking is widely acknowledged in visualization literature. Cleaveland noted that in
“elementary graphical-perception tasks” the preattentive perception of basic graphical
elements underlies data visualization (9). Abarbanel suggests that visualization can
be defined as the substitution of “preconscious visual competencies” for “conscious
thinking”(10). Woods indicated that “If the mental activities required to locate base
data units are not automatic, but require mental effort or capacity, they may disrupt or
divert.. .the main line of reasoning” (11).
2.5 Conclusion
Intelligence plays a vital role in the decision making process both in and out
of a military conflict. Reaching these decisions is no small task. Navigating the sea
of information through our vast data repositories is a large undertaking. Confronting
this intelligence problem is Project Broadsword. While making great strides in the
area of data management. Project Broadsword is also investigating more advanced
data exploitation methods through visual cues. The field of data visualization, though
constrained by technology and human factors, is an expanse of possibilities waiting to
18
be explored. Exploration of the visual realm leads immediately to the use of color.
Association through color is an inherent human trait that is very fast, efficient, and
effortless. This research embraces that fact and exploits it.
19
3. Methodology
3.1 Data Exploration
This research provides an effective representative environment for
exploitation of stored military intelligence. The philosophy of design is to provide
the intelligence analyst with enhanced visual cues and the spatial freedom to explore,
analyze, and evaluate products from a chosen geographical area of interest in three-
dimensional space. The Project Broadsword Test and Evaluation team conducted
intelligence product exploitation tests with and without the use of data visualization
techniques. Their subjective comparisons between data visualization and a text-only
approach lay the groundwork for the findings in Chapter 6 and the conclusions and
recommendations in Chapter 7.
3.2 Software Development Process
The nature of the Project Broadsword data visualization problem lends itself
nicely to the construction of a partial implementation, or prototyping. Within
prototyping there are two schools of thought, the throwaway approach and the
evolutionary approach (12). The throwaway approach is concerned with the
construction of software in order to learn more about the problem or its solution.
Once this is understood, the prototype is completely discarded. The evolutionary
approach also examines the problem or its solution; however, once the requisite
knowledge has been gained, the prototype is readapted and used again. This process
repeats itself until the prototype eventually becomes the actual system.
The intelligence community is beginning to explore how the use of data
visualization can satisfy its users’ needs. Responding to this trend. Project
20
Broadsword is emphasizing the “look and feel” of a final data visualization design,
not its underlying implementation. This research provides the means by which an
analyst can explore and adapt to new data visualization ideas and concepts. For this
reason, I have chosen throwaway prototyping as the software development process.
3.3 Visualization Considerations
Applying methods of good data visualization techniques can be difficult.
Although a significant amount of background work goes into producing “good” data
visualizations, a large part of this thesis relies on visual experimentation and the
associated human response. Regardless of technical orientation, each Project
Broadsword tester’s feedback is partially subjective and based on their personal tastes
and prior data visualization experience. Final analysis of this research and the
associated conclusions must take this subjectivity into account. This research uses
technically sound approaches to explore appealing graphical representations.
Technically perfect data visualizations will not be utilized if they do not appeal to the
user.
As stated in Section 2.4.1, color is a major factor in the success or failure of
data visualizations. For this reason, color is heavily examined throughout this
research. Evaluating the effects of different colors at varying levels of resolution and
intensity account for a significant amount of effort.
Transparency is a unique graphics capability that, when used effectively,
allows for the interior of a three-dimensional object to be used as additional data
presentation space. This research explores the additional capabilities provided by this
method of object creation.
21
3.4 Data Navigation
Intelligence products are constructed with more data elements than can be
comfortably mapped to the visual realm. Therefore, a successful visualization is
dependent on the environment and its ability to handle this overload. When viewed
from this perspective, data visualization takes a back seat to navigation and is handled
accordingly throughout this research.
3.5 Scenario Generation
To validate the results of this research, a scenario was constructed which
simulates real-world intelligence product exploitation. This scenario, seen in Figure
3.1, exercises the current Project Broadsword textual and two-dimensional product
retrieval functions. The scenario is then exercised using the newly developed three-
dimensional data visualization methods.
For each experiment, testers are provided with the result formats and tasks as
outlined in Figure 3.1. Providing the results eliminates the need for testers to perform
the actual searches and ensures each tester has an identical display at the start of each
test. Following each test, feedback is generated through tester interviews. Each tester
subjectively rates the experience for ease of use, suitability to task and perceived
performance. Findings and conclusions based on a comparison of test results are
found in Chapters 6 and 7.
The scenario in figure 3.1, used for testing purposes, is constructed of
intelligence products from three intelligence sources (the Military Intelligence
Database (MIDB), the Image Product Library/Image Product Archive (IPL/IPA) and
22
the Automated Message Handling System (AMHS)) and contains all four product
types; i.e. imagery, video, text and audio.
Broadsword Geospatially Oriented Navigational Environment
Experiment Scenario
Instructions: Given a set of one hundred and seventeen
intelligence products and three result formats, complete the
three tasks in any order. Utilize the accompanying color
identification map for product-to-glyph color associations.
Result Formats: 1. Text-only
2. Two-dimensional map with icons
3. Three-dimensional map with glyphs
Task #1: Locate the total number of textual products
available.
Task #2: Retrieve all imagery products that have IPL as
a data source.
Task #3: Identify all F-15E aircraft imagery
products.
Intelligence Product-to-Glyph Color Identification Mapping
Color
Product Type
Green
Imagery
Purple
Textual
Red
Video
Yellow
Audio
Gray
Multiple Products
Figure 3.1: Experiment Scenario
23
4. Development Environment
4.1 Environment
To confront the growing stream of intelligence data, as stated in Section 1.3,
while meeting the challenges of data management as outlined in Section 1.4, this
research effort created an environment to explore proof-of-concept data visualization
capabilities tailored to the Project Broadsword intelligence data retrieval and
exploitation system. This environment, the Broadsword Geospatially Oriented
Navigational Environment (BGONE), was constructed in accordance with the Project
Broadsword System Security Requirements and Analysis document. This document
provides the vehicle for listing, and establishing compliance with, the minimum
technical and nontechnical requirements for the Project Broadsword System in
processing U.S. intelligence in the System High mode (classified mode) of operation.
Additionally, it provides an analysis of Project Broadsword’s safeguards intended to
satisfy system security requirements where applicable.
Due to the sensitive nature of Project Broadsword’s mission, BGONE is
constrained to coding languages and techniques authorized by the Project Broadsword
program office. These constraints constricted BGONE’s evolution by significantly
reducing the number of possible development paths.
4.2 Target Environment
Although BGONE is a stand-alone environment, its implementation mirrors
the functional design of Project Broadsword. A review of available documentation
and operational parameters revealed that Project Broadsword is deeply committed to
the cross platform belief. It utilizes only those processes and techniques to interact
24
with the system from virtually any platform. Additionally, the system is carefully
tailored to include only those programming languages that support seamless operation
through an Internet Protocol network. These languages include the Hypertext
Markup Language (HTML), JavaScript, the Common Gateway Interface (CGI) and
the Virtual Reality Modeling Language (VRML). Through strict adherence to these
developmental policies, users are able to perform their duties by simply utilizing a
standard World Wide Web (WWW) browser such as Internet Explorer or Netscape
Navigator. To ensure accurate results, BGONE conforms to this operating
environment paradigm.
4.3 Project Broadsword Code Base
Project Broadsword is currently undergoing a major code rewrite. Of key
importance to this research is the map server and its associated data handling
functions. As these are being re-coded from the ground up, their completion times
exceed that allotted for this study. To support this research without having a fully
functional Project Broadsword map server, BGONE was created as a stand-alone
entity. The new map server relies heavily on VRML and JavaScript code; therefore
these languages make up BGONE’s core code base.
4.4 Model Approach
BGONE provides exploration of intelligence data handling concepts through
the use and manipulation of three-dimensional VRML models. BGONE’s primary
focus is the representation of the data, and not the terrain over which it is displayed.
Therefore, three-dimensional maps are externally generated and imported as needed.
The Project Broadsword development team provided map models as needed.
25
4.5 Evaluation of Hardware Platforms
Multiple hardware platforms are available within the Project Broadsword
development lab. These platforms include SUN, Silicon Graphics, and x86 based
machines of varying speeds and capabilities. Based on suitability to task, ease of use,
and support for VRML, the x86 architecture was chosen as the development platform.
Although SUN systems are prevalent in the lab, UNIX based platforms do not yet
fully support the VRML language.
4.6 Software Development Languages
BGONE was developed using JavaScript, HTML, and VRML. These
languages were pre-determined by the Project Broadsword project office so no
evaluation of alternate tools was performed. VRML was chosen as the only available
three-dimensional graphics language that adheres to Project Broadsword’s strict
security, integrity and IP requirements. VRML provides a robust graphics-rendering
library but lacks functionality in the area of data navigation. JavaScript and HTML
fill the void by providing effective data selection and interaction.
BGONE is primarily based on the 1997 International Organization for
Standardization (ISO)/Intemational Electrotechnical Commission (IEC) document
14772, Virtual Reality Modeling Language (13). All VRML code is constructed
using this document as a guide. Additionally, the JavaScript and HTML support code
are integrated using established industry practices.
4.7 Automated Tools
To facilitate the development of BGONE, I employed the use of Spazz3D by
Virtock Technologies, a powerful VRML object generation tool. Spazz3D provides
26
key functions which include:
• A GUI based interface to accommodate simple geometry creation by a
visual drag-and-drop method
• Software wizards that aid complex tasks such as creating animations and
moving cameras.
• OpenGL support for enhanced graphics rendering
• Easily implemented directional lights, point lights, and spot lights
While Spazz3D provided the bulk of objects used in BGONE, assembling the
environment and employing data navigation methods required coding by hand. This
was accomplished with PC Editor, a simple multi-windowed ASCII text editor.
4.8 Graphics Rendering
BGONE graphical rendering is accomplished using OpenGL. Developed by
Silicon Graphics Incorporated in 1992, OpenGL is the most widely adopted graphics
standard in use today and is supported on virtually all platforms. This fact coupled
with its ability to produce visually consistent displays, regardless of operating system
or windowing system, maintains Project Broadswords’ goal of universal functionality
and presentation across heterogeneous platforms. When OpenGL compliant
hardware is unavailable or not configured, software emulation is seamlessly enabled.
4.9 External Support
BGONE, developed with interpretive based languages, requires a JavaScript
enabled WWW browser with an associative VRML plug-in to function properly.
Since no web browser supports VRML code directly, a helper program or plug-in is
required. Although the code in BGONE is based on VRML97, JavaScript 1.2, and
27
HTML 3.2 standards, each browser supports a particular standard with varying
degrees of success. Due to their lack of compliance and poor documentation on the
part of VRML, JavaScript and HTML; the success or failure of a specific design can
only be validated following completion. The testing of the various combinations of
web browsers and plugins played a crucial role in BGONE development and is
explained in section 4.10.
4.10 Commercial Off The Shelf Products
Project Broadsword prides itself in leveraging off of existing technologies.
Due to its web design, it relies on commercially based WWW browsers and VRML
plugins. There are three major players with respect to browsers: Internet Explorer,
Netscape, and Opera. Internet Explorer is a Microsoft standard browser and is
available for the x86 Windows and Sun Solaris based architectures. Netscape caters
to virtually every architecture. Opera is a x86-based browser that is under
development. Of these three, Netscape is the primary browser used by the Project
Broadsword development team as it has the most support. Based on this fact,
Netscape was also chosen as the web viewer for BGONE.
As inferred in section 4.9, Netscape does not support a VRML environment
directly. It requires an external plugin to perform this function. Three VRML
plugins were considered for use: Cosmo, Blaxxun Contact 3D and Cortona. Cosmo
has been a leader in VRML technology in the past but development has been halted
and it is no longer supported. Blaxxun Contact 3D and Cortona are currently the only
two actively supported VRML plugins. Both take advantage of OpenGL accelerated
hardware while providing adequate navigational functions. The only true
28
differentiation is the amount of real estate used by the on-screen VRML controls.
Blaxxun Contact 3D excels in this area; utilizing a relatively small footprint and thus
was chosen as the BGONE companion to Netscape.
29
5. Design and Implementation
5.1 Environment
The Broadsword Geospatially Oriented Navigational Environment (BGONE)
is a stand-alone software environment that provides an effective representative
atmosphere for exploitation of stored military intelligence. Modeled for Project
Broadsword, BGONE is a stand-alone environment that provides an effective
representative atmosphere for exploitation of stored military intelligence. BGONE
design and implementation is based on design concepts of the new Project
Broadsword 3.x map server. It also complies with policies and procedures congruent
with mandated security requirements outlined in Section 4.2.
BGONE provides a one-to-one mapping of intelligence data product metadata
elements, such as product type, to a specific visual attribute like color. Mapping is
provided by pre-selected user input and is a function of the environment architecture.
Project Broadsword prides itself in offering the user a plethora of configurable
options when using the system. Likewise, the integration of concepts and techniques
developed in BGONE are intended to be user configurable via the Project
Broadsword user preference interface. Hence, each individual user can select which
visual attribute maps to which data element to satisfy their specific needs.
The most important contribution of BGONE is the ability to map intelligence
data, not decide what the ideal mapping should be. For the purposes of this research,
Table 5.1 depicts the mapping of color to product type, i.e. imagery, video, text or
sound. Additionally, the use of the color gray designates that multiple products are
present but not shown. Use of this particular mapping along with others will be
30
further discussed throughout this chapter.
Table 5.1: Intelligence Data Mapping
Product Type
Visual Attribute - Color
Imagery
Green
Textual
Purple
Video
Red
Audio
Yellow
5.2 Three-dimensional Visualization
User feedback obtained through informal interviews revealed a need for an
environment where intelligence products could be exploited through visual means.
Leveraging off of the old adage that “a picture is worth a thousand words”, user’s
expressed the desire for a mechanism that provided increased understanding of a
given set of intelligence products within a narrow window of time. Specifically
implicated was the additional possibilities generated by the use of the third
dimension. Adding this additional dimension significantly increases the amount of
usable virtual real estate for data presentation.
Responding to user feedback, this research did not pursue two-dimensional
data visualization. Instead, BGONE moves directly from a textual to a three-
dimensional representation. This transition poses technical obstacles that are a non¬
issue in two dimensions. These obstacles include occlusion as objects are placed
directly in front of each other, user perspective at a given instance in time, and data
navigation.
5.2.1 Three-dimensional Objects
Managing representative objects in data visualization begins with the creation
of the objects themselves. To accommodate BGONE development a glyph-based
31
visualization approach was employed. A glyph is an icon or graphical object that is
affected by its input data and certain properties such as x, y, z locations in three-
dimensional space, color, opacity, orientation, shape and size.
Glyphs differ from standard graphical objects because they provide a direct
one-to-one mapping between a selected data element and a glyph attribute. For
example, an intelligence product type can be directly mapped to the color attribute.
Using the mapping of table 5.1, figure 5.1 symbolizes four imagery products. Since
figure 5.1 displays four unique shapes, a second data element, such as the branch of
military service that produced the imagery, could be mapped to the shape attribute.
This second mapping would increase the usefulness of the data visualization by
Figure 5.1: Symbolic Glyphs (Color Plate 3)
32
helping the analyst visually segregate the search space by product and image creation
source. As more ‘intelligence data element to glyph attribute’ mappings are
performed, the given search space is further populated with more graphically
represented product details. Each added representation increases the analysts’ ability
to quickly select specific products of interests without immediate textual intervention.
For BGONE, initial prototype glyphs were created using Spazz-3D, which is
described in section 4.10 Automated Tools. Using Spazz-3D it was possible to
quickly create glyphs, such as those seen in figure 5.1, for immediate examination.
5.2.2 Multilevel Glyphs
Users specifically requested three-dimensional data visualization. Their
request was motivated by the knowledge that using the third dimension allows for the
use of depth, which provides more usable space for data presentation. To ensure
maximum screen space usage, glyphs were created at two levels of data resolution;
level one and level two. A level one glyph, often referred to as the outer glyph, is the
primary glyph and provides identification of individual products such as imagery.
Each level one glyph represents one intelligence product. A level two glyph, often
referred to as the inner glyph, is the secondary glyph and provides specific
information about that particular product. Figure 5.2 shows representation of two
products. Based on this example, the green level one glyphs symbolize two imagery
products. The level two glyph, which can be seen inside the level one glyph, can now
further the analysts’ understanding of the product through a one-to-one matching of
product element to glyph attribute. Here a cube may represent ground-based imagery
while the cone may represent air-based imagery. Additionally, the color of the inner
33
glyph offers another mapping level such as product origin. As shown, this glyph in a
glyph visual approach helps maximize the useable space taken up by each product
representation. Although this does not effect the number of products displayed, it
does increase the amount of information portrayed.
Figure 5.2: Multilevel Glyphs (Color Plate 4)
5.2.2.1 Level One Glyph
To allow the analyst to identify each intelligence product, a level one glyph
represents one product in its entirety. The first area of interest dealt with the shape
used when displaying this glyph. The use of shape as a data-mapping attribute
required careful examination. Initially, each of the four product types was mapped to
one of the geometric shapes shown in figure 5.1. Preliminary testing indicated that
the vertical stacking of these glyphs in relatively small concentrations (20 products or
34
less) resulted in an acceptable representation. However, when the number of products
displayed increased, the virtual landscape began to take on a cluttered appearance.
After developing several models and consulting with the Project Broadsword program
office, it was decided to maintain a uniform shape for all outer glyphs.
Four common geometric shapes were candidates for the level one glyph: the
cube, the sphere, the cone and the cylinder. Each was evaluated for its appearance
after being vertically stacked and placed side-by-side. Based on the subjective
observation that they do not stack well and wasted the most space per object, the cone
and sphere were eliminated as contenders. As can be seen in figures 5.3 and 5.4, the
two remaining choices offer a uniform eye appealing presentation. The final decision
Figure 5.3: Cylinder Shaped Level One Glyphs (Color Plate 5)
35
to use a cube-based glyph was made to reduce the number of polygons required to
generate each glyph.
Figure 5.4: Cube Shaped Level One Glyphs (Color Plate 6)
With the cube selected as the level one glyph, attention was turned to the use
of color. As explained in section 2.4.1, color is a very effective data visualization
tool. For the purpose of this research, the color of the outer glyph represents a
specific product type. Table 5.1 provides the color mapping used.
Exploiting the inner glyph only succeeds if the outer glyph provides visual
access to its interior. One approach is to use transparency. Transparency provides
interior access while maintaining the outer glyphs shape and color integrity. This
technique also allows a user to see other products in the distance without changing
perspective. Figures 5.5 and 5.6 examine varying levels of transparency and the
36
Figure 5.5: Transparency as Designed (Color Plate 7)
ability to distinguish the inner glyph while maintaining the outer glyph. Figure 5.5
displays transparency as viewed with the Spazz-3D, a VRML object generation tool.
Using Netscape and its associated VRML plugin Blaxxun Contact 3D, as required by
Project Broadsword, resulted in figure 5.6. Examining these two figures highlights
two major inconsistencies. First, figure 5.5 demonstrates six levels of transparency as
designed. Notice that at each level, both the inner and outer glyph is discernable.
Figure 5.6 shows the same six glyphs, as viewed in an operational environment. It is
clearly evident that as the level of transparency increases the visibility of the outer
glyph diminishes until it is no longer discernable.
Another adverse inconsistency is the textural appearance of the outer glyph.
The original design, figure 5.5, was intended to provide a smooth fog like effect, not a
coarse texture. The coarseness displayed in the top row of figure 5.6 is easily
mistaken as intended texture. Although mapping intelligence data to texture may lead
to more data represented, this was not the intent and it visually detracts from the inner
glyph. This detraction reduces the inner glyphs ability to effectively provide data
visualization mappings by making it difficult to see, thus reducing its utility. Being in
three-dimensional space, the coarseness also reduces the ability to see objects directly
behind the glyph being viewed.
The diminished visual quality, as is evident by the lack of definition and
coarseness in figure 5.6, severely impacted the decision to pursue transparency. The
dismal results of transparency further emphasized the lack of VRML compatibility
from one VRML interpreter to another.
38
A second, more successful approach to outer glyph creation is shown in figure
5.7. The outer glyph is constructed of twelve smaller three-dimensional boxes that
Figure 5.7: Redesigned Level One Glyph (Color Plate 9)
are arranged to form the original cube. This layout maintains a uniform color mapped
cube while providing easy visual access to its interior. An added feature of this
scheme is that the transparent interior does not alter the appearance or color of the
inner glyph.
5.2.2.2 Level Two Glyph
Level two glyph development initially progressed as a mirror of level one.
Results of this are shown in figure 5.8. As shown here with the outer green cube
symbolizing an image product, the inner glyph attributes of shape and color are
39
Figure 5.8: Abstract Level Two Glyph (Color Plate 10)
available for mapping two additional product elements. Collaborative meetings with
the Project Broadsword program office revealed a desire for a more gradual
introduction of data visualization with a smaller learning curve. Figure 5.8 requires
the user to have a previous knowledge of the inner glyph mappings. A more intuitive
approach is to integrate a VRML model of the product type being represented. This
is demonstrated in figure 5.9 with an imagery product of an aircraft.
5.2.3 Level Of Detail
To minimize visual clutter when viewing a large number of products from a
distance, a Level Of Detail (LOD) scheme was employed in two stages. First, low
LOD models were created to give analysts a visual understanding of the level one
glyph without requiring knowledge of the level two glyph. These models, as seen in
40
Figure 5.9: Model Based Level Two Glyph (Color Plate 11)
figure 5.10, render only the outer glyph. Masking the inner glyph permits the outer
glyph to be rendered as a solid cube. A solid cube provides more surface area for the
color attribute, as compared to figure 5.7, allowing the viewer to more easily identify
Figure 5.10: Low Level Of Detail Glyphs (Color Plate 12)
41
a specific glyph from a distance. When viewing data in a three-dimensional
landscape, the ability of the user to see all rendered elements is crucial.
Users select a key product element to be mapped to the color of a level one
glyph. Their selection allows them to quickly locate products of interest based on the
task at hand. For example, an analyst may be directed to analyze all available
imagery over a specific geographic area. If the inquiry returns a mix of products, low
LOD models with a color attribute of green will be readily evident. As seen in figure
5.11, low LOD models used with a VRML landscape map provide immediate
feedback without the distraction of individual level two models.
Figure 5.11: Low Level Of Detail Glyphs on a VRML Landscape Map (Color Plate 13)
Low LOD models minimize visual clutter by using proximity sensors. As an
analyst moves closer to glyphs of interest, the models change to that seen in figure
5.12, revealing additional information about the products. When the analyst is
42
finished in one location and moves towards another, the first glyphs revert to their
original form as shown in figure 5.10.
Figure 5.12: Proximity Sensor Activation Rendering (Color Plate 14)
Level Of Detail is also concerned with the number of products displayed at a
given time. Even with three dimensions, display space is a limited commodity and
must be managed accordingly. Project Broadsword testers frequently perform
searches that result in over eight hundred products. To accommodate such a large
number, the gray glyph at the bottom of Figure 5.12 was introduced to signify that
multiple products are present but not visible.
For this research. Figure 5.10 or Figure 5.12 provides an understanding of
seven products; textual, imagery, and five that require additional actions to explore.
43
A gray glyph is explored by selecting it with the mouse and clicking a mouse button.
A second web browser window is then generated, as seen in figure 5.13, which
displays the products found at that location.
Figure 5.13: Products Represented by Gray Glyph (Color Plate 15)
5.3 Navigation
When dealing with a two-dimensional plane, navigation is primarily
concerned with how the eye traverses the scene. This is controlled through effective
use and placement of lines, shapes and color. Entering the third dimension introduces
an expanded domain in which the user must physically navigate the data.
Individual intelligence products provide a large amount of data and therefore a
one-to-one matching of each data element to a visual cue is unrealistic. Key product
44
metadata elements are therefore used for visual rendering leaving the remainder for
textual displays. This allows for quick visual selection of a desired product. This
separation in data and metadata necessitates the need for two types of navigation:
visual environment navigation and data navigation.
5.3.1 Graphics Navigation
Visual environment navigation is provided through Blaxxun 3D as discussed
in section 4.10. Table 5.2 provides a cursory navigation overview. Particular
keyboard and mouse actions for each of the stated movements can be found in the
Blaxxun 3D help menu.
Table 5.2: VRML Navigation
Walk
Allows the user to move through the scene
in an X and Z direction
Slide
Allows the user to move through the scene
in an X and Y direction
Examine
Allows an entire scene to be rotated around
its own X axis
Rotate
Allows an entire scene to be rotated around
its own X and Z axis
Fly
Allows the user to move through the scene
in an X, Y, and Z direction.
Pan
Allows the entire scene to be rotated
around the users X and Y axis
Additional movement through the use of coded Viewpoints is also provided.
Viewpoints allow an analyst to quickly move to an area of interest with a click of the
mouse or a limited sequence of keystrokes. A segment of Viewpoint code is shown
here:
45
DEF ViewPointl Viewpoint {
description "ViewPointl"
jump TRUE
fieldOJView 0.79
position 0.0 0.0 0.0
}
This segment defines an X, Y, Z position relative to the current coordinate system and
a field of view, in radians, indicating the spread angle of the viewpoint’s viewing-
volume frustum. Additionally, the jump being set to TRUE ensures the new
viewpoint becomes the current user view.
The Blaxxun 3D controls coupled with Viewpoints allow a user to gracefully
move through BGONE, observing the various products and data elements presented.
However, navigation of the visual environment provides only a partial solution. Text
handling must also be integrated.
5.3.2 Textual Navigation
Once a desired product is visually selected for exploitation, the user must have
effective text handling techniques to benefit from the initial data visualization. As
seen in figure 5.14, integrating a text box with a VRML environment is conducive to
the use of frames. Project Broadsword already utilizes frames so their inclusion was
seamless.
Text handling through the use of frames allows the analyst to maintain the
visual results while examining the textual details. This is an improvement over a pure
text approach, but more advanced methods through dialog boxes were explored.
Natively, the VRML language provides a function termed TouchSensor that
enables animation based on mouse interactions. For example, placing a mouse cursor
46
e Edit Yw &u Comnnunrcalor yelp
4 * 1 $ j?»- jal -i sf
Reload Home Search Netscape Seamy
^'Bookmorfc* & LocoJion: Itile ///q/Mrchee! GfTbesis Model/BGONE hfrn
over a sphere and pressing the left mouse button can cause the sphere to spin on its
axis. This interaction is an ideal candidate for the support of pop-up dialog boxes
containing intelligence product data. Examination and experimentation with these
touchsensors revealed VRML is not capable of providing context-sensitive dialog
boxes. A supporting language, such as JavaScript, was required. JavaScript is well
situated for browser-based text handling tasks. However, being designed to co¬
habitat with HTML, it does not readily support the VRML environment.
47
For this research, methods were created for the control and generation of data
elements when interacting with glyphs. These new methods combined the animation
functions of VRML with the text handling capabilities of JavaScript. The resulting
functionality enables BGONE to meet the needs of Project Broadsword by providing
quick and efficient access to data elements not mapped to a glyph attributes.
These new methods provide context-sensitive dialog boxes and data handling,
through three VRML interactions: isActive, isPressed and isOver. Figure 5.15 shows
^Additional Information - Netscape_HEP
* Keymaster: NOSC
* Platform: 11450
* POC: Colonel Road
•DSN: 666-1111
Figure 5.15: Pop-up Dialog Box
a resulting window when the isActive method is called. This method opens a dialog
box whenever a mouse button is pressed and held while over a glyph. When the
mouse button is released the dialog box closes. Figure 5.16 demonstrates the
isPressed method that generates the associate data, in this case an F-15E, in the right
frame of the browser window. This is accomplished by placing the mouse cursor
over a glyph and single clicking the mouse. Figure 5.16 also shows activation of the
isOver method. This method opens a dialog box whenever the mouse cursor is
moved over a glyph. The box remains open until the cursor is moved completely off
of the glyph. The automatic opening and closing of the dialog box provides a very
quick glance capability as the mouse cursor is moved from one product to the next.
48
|Pnr* S*curty Shop
59
3 _
Dialog box generated
by the isOver method
F-15E Strike Eagle
Mission
The F-15E Strike Eagle is a dual-role fighter designed to perform
air-to-air and air-to-ground missions. An array of avionics and
electronics systems gives the F-15E the capability to fight at low
altitude, day or night, and in inclement weather and perform its
primary function as an air-to-ground attack aircraft
Features
The aircraft uses two crew members, a pilot and a weapon
systems officer. Previous models of the F-15 are assigned
air-to-air roles; the "E" model is a dual-role fighter. It has the
capability to fight its way to a target over long ranges, destroy
enemy ground positions and fight its way out
An inertial navigation system uses a laser gyro to continuously
monitor the aircraft’s position and provide information to the
central computer and other systems, including a digital moving
map in both cockpits.
The APG-70 radar system allows air crews to detect ground
e. — 1.—-A, ry — w.-. _o._ J
" . • -•& 'Vj, 'ip £3
Data placed in frame via
the isPressed method.
Figure 5.16: Text Handling (Color Plate 17)
49
5.4 Multimedia
To expand the data exploitation capabilities within BGONE, multimedia
concepts were explored. Specifically targeted was the integration of streaming video.
To enhance exploitation of this medium, the inner glyph was cast as a cube with the
video stream texture mapped onto it. Video support is provided through the use of the
Real Audio player plugin. Playing of the video is triggered by activation of a
touchsensor.
5.5 VRML Plugin Verification
Operation of BGONE requires a VRML plugin be present. Detection
software was created to prevent program execution if a plugin is not installed. This
new code identifies which browser is being used and then checks for known x86
based VRML plugins. These include Blaxxun 3D, Contact, Cosmo Player, Microsoft
VRML 2.0 Viewer, Viscape Universal and Worldview. If a plugin is detected the
user notices no difference during startup. If not, an informational message is
displayed.
5.6 Conclusion
This chapter focuses on the design approach and obstacles encountered during
the implementation of BGONE. It was clearly evident early on in the research that
design tradeoffs would have to be made in order to mirror the security and integrity
mandates placed on Project Broadsword. The creation of new, never before seen,
data handling methods gives BGONE a unique operating base on which to explore
further data visualization and text handling capabilities.
50
6. Findings
6.1 Introduction
The primary objective of this thesis is to show that three-dimensional data
visualization can provide an analyst a more efficient and effective way of organizing,
understanding and navigating vast amounts of intelligence products. This
visualization approach is specifically tailored to Project Broadsword, an intelligence
data retrieval system, being developed at the Air Force Research Laboratory/Rome
Research Site in Rome, New York. The creation of the Broadsword Geospatial
Oriented Navigational Environment (BGONE), a stand-alone representative
environment of Project Broadsword, provided a tightly controlled test environment
through which the Project Broadsword verification and validation team could
compare their current intelligence exploitation methods to those provided by
BGONE. The findings presented here are based on results while employing the test
scenario outlined in Section 3.5. For the remainder of this chapter, the Project
Broadsword verification and validation team will be referred to as ‘testers’.
6.2 Experiment Setup
Intelligence products used in this research were obtained from unclassified
sources located inside the Project Broadsword development, testing and integration
vault. Three sources were utilized; the Military Intelligence Database (MIDB), the
Automated Message Handling System (AMHS), and the Image Product
Library/Image Product Archive (IPL/IPA). These sources provided products from all
four types; audio, video, imagery, and text documents.
51
Following the guidelines presented in Section 3.5, testers performed each task
three times. First, each scenario was completed using the live Project Broadsword
system in ‘text results only’ mode. Next, the scenario was accomplished with the
system in ‘geospatial map results’ mode. Finally, the scenario was exercised using
BGONE.
6.3 Experiment Results
Experiment results are based on the combined responses received during tester
interviews upon completion of each scenario. Each tester completed a single
experiment cycle (the scenario exercised in each of the three configurations as
described above) once.
6.3.1 Visual Presentation
People have a natural acceptance and appeal for visual stimuli. As discussed
in Chapter 2, the average individual can recognize and comprehend a complete
graphical image much faster than reading a single line of text. BGONE taps into this
acceptance by providing an environment where cognitive functions are stimulated
through visual depictions of intelligence products. Embracing this phenomenon,
testers found themselves naturally drawn to the scenario results provided by BGONE.
Although not a technical measure, they found the rich color and sense of depth caused
by using three-dimensional objects pleasing to the eye. Compared to the existing
two-dimensional approach, testers felt that BGONE portrayed an enhanced realism
that allowed them to categorize and explore geographically based intelligence data
more quickly. The most prominent factor to this perceived speedup was the testers’
ability to more easily comprehend the overall clustering of products within the area of
52
interest. Testers also commented that the ability to alter one’s viewpoint without
redrawing the scene, as is currently done in the two-dimensional geospatial map,
allowed for a more fluid transition when directing attention from one product to the
next. In response to the added visual intensity of utilizing three-dimensional space,
testers overwhelmingly conveyed that its use enhanced the data exploitation
experience. Their observations were based on the fact that the increase in the number
of products able to be viewed at a single time is a vast improvement over the current
two-dimensional map and textual based display. The ability to view a large number
of products in a given instance of time increases efficiency by reducing the amount of
time and user intervention required to browse through all products of interest.
Besides visual acceptance, testers examined whether BGONE increased
understanding of the total number of products available when compared to the two-
dimensional map. Testers found that BGONE increased product results knowledge
by enabling a better understanding of the number of products located at a given
location. This is accomplished in two ways. First, each individual product is
represented and displayed as a single glyph. However, when the number of products
available at single location exceeds a user-specified parameter or the available scene
space is depleted, a gray glyph takes on the representation of multiple products. For
example, with a gray glyph set to represent five products, Figure 6.1 displays seven
available products.
The two-dimensional map, seen in Figure 6.2, uses a single green square to
represent all products at a given location. There is no mechanism to enable the user
to gain an understanding of the number or types of products available. This
53
Figure 6.1: Glyph Representation (Color Plate 18)
Single Green Square
Figure 6.2: Two-dimensional Geospatial Map (Color Plate 19)
understanding requires the icons to first be explored. Exploring an icon requires the
opening of a new window and returning to a text based format. Testers found this
method of operation to be counter productive.
Beyond the number of products, testers preferred the BGONE method of
mapping product types to the color of the outer glyph. When performing scenario
task one, the ability to quickly identify textual products by the color purple made for
light work. Presented with the two-dimensional map, task one required the
exploration of every icon coupled with a text based visual search for products marked
as ‘text’. Using BGONE, gray glyphs also had to be explored; however, they did not
require textual searching. Overall each tester performed task one in approximately
fifty percent less time when utilizing BGONE.
6.3.2 Visual Navigation
Three-dimensional data visualization over a geospatial map may require a user
to physically navigate the virtual space. Since task one, locating the total number of
textual products, only required identification of the outer glyph, it could be
accomplished using only low level of detail glyphs (as described in Section 5.2.3).
Glyphs can be selected at any visible distance, therefore, no movement within the
virtual world was required. Task two, retrieving all imagery products that have an
IPL source, forced testers to move about the virtual landscape tripping proximity
sensor to reveal the inner model. Testers found this movement difficult and time
consuming. The most common problem was loosing track of their current virtual
location. Additionally, maintaining the correct visual perspective required careful
movements to prevent disorientation. When compared with the textual and two-
55
dimensional map based approach, performing task two with BGONE took
approximately four times longer to complete. Prior to this experiment, most users
had little to no experience navigating in a three-dimensional environment.
6.3.3 Textual Data Navigation
Task three, accounting for all F15-E imagery products, built on the
requirements of task two by adding the additional need to locate more data than can
be visually displayed. Testers had the same navigational experiences as described in
Section 6.3.2, however, once there they had to contend with locating the correct
information. When fulfilling task three using the textual and two-dimensional map
approach, testers searched through a laundry list of products to locate the required F-
15E imagery. As shown in Figure 6.3, using frames and automatic pop-up dialog
boxes, as described in Chapter 5, BGONE enabled testers to more efficiently locate
the needed information. This is made possible because a dialog box provides a
quick-look capability for key elements of information. Figure 6.3, therefore, enables
us to quickly view the product type, resolution of the image, the data source, and the
originating gatekeeper without having to sift through all the data available. Although
the testers highly praised the text handling of BGONE, getting to that point in the
experiment caused task three to take five times longer to complete versus using the
textual or two-dimensional map methods.
56
ttf "" Document Done
gStartj JOemc3 j Jhogeaot* j JKfahadG j JkjGAggjicJ .jTtesis Model] jAugns | SOtoeen ca... \ jjMyCompi*;t| _j(Cj
F-15E Strike Eagle
Mission
The F-15E Strike Eagle is a dual-role fighter designed to perform
air-to-air and air-to-ground missions. An array of avionics and
electronics systems gives the F-15E die capability to fight at low
altitude, day or night, and in inclement weather and perform its
primary function as an air-to-ground attack aircraft
Features
Ik aircraft uses two crew members, a pilot and a weapon
systems officer. Previous models of the F-15 are assigned
air-to-air roles; the "E" model is a dual-role fighter. It has the
capability to fight its way to a target over long ranges, destroy
enemy ground positions and fight its way out
An inertial navigation system uses a laser gyro to continuously
monitor the aircraft’s position and provide information to the
central computer and other systems, including a digital moving
map in both cockpits.
The APG-70 radar system allows air crews to detect ground
*_ _ c. i- - a _. .r<i:-*— u.-» -a.- ZJ
.. ' \,4l -P .(3
| Opera 3.60 (-1 ^Broadsword I jjgProductA.. 5:24AM
Figure 6.3: Product Exploitation with BGONE
57
6.4 Conclusion
These findings paint a mixed picture, depending on which issues you place the
most emphasis. Data visualization in three-dimensional space appears to bring new
opportunities with challenges to match. Despite the challenges, testers found
BGONE to be a solid basis for which to test and examine Project Broadsword data
visualization issues. Conclusions and recommendations for further research based on
these findings are presented in Chapter 7.
58
7. Conclusions and Recommendations
7.1 Introduction
This research set out to explore how data visualization techniques could
enhance an intelligence analyst’s exploitation capability. Specifically targeted were
Project Broadsword and its unique ability to access and disseminate data from
heterogeneous sources worldwide. To investigate data visualization concepts in
controlled experiments required construction of a representative modeled
environment. This environment, known as the Broadsword Geospatially Oriented
Navigational Environment (BGONE), supplied the framework and tools to simulate
how an adept user in the field would respond and navigate a geospatially disperse
field of data.
This chapter discusses implementation considerations based on the findings in
Chapter 6. Unique contributions to the field, developed as a result of this research,
are also covered. Finally, recommendations for future research are presented.
7.2 Conclusions
The experiments run in BGONE offered a wealth of information. The most
notable being that the primary concern with data visualization in a three-dimensional
space is not the visuals, but rather the navigation required moving from one product
to the next. Given that our dataset is geospatially placed on a worldwide scale, the
amount of virtual terrain that must be traversed may be quite large. Using only
standard two-dimensional input devices, a mouse and keyboard, most testers who had
to travel more than a few virtual meters found the experience quite frustrating.
The intelligence data products used for this research contain a large number of
59
data elements. This large quantity is more than purely visual methods can effectively
display. The addition of pop-up dialog boxes effectively aided testers in the
completion of experiment tasks requiring exploitation of specific data elements held
by a product. Unlike the two-dimensional map method, testers were able to quickly
move from one product to the next without returning to a text-based laundry list of
products.
This research focused around the display of representative geometric objects
in three-dimensional space. Data visualization is a powerful tool that becomes
exponentially more complicated to develop and control when a user must physically
interact with the data they are viewing. Those who have little to no experience with
three-dimensional environments may find this type of data visualization
uncomfortable to work with. As discussed in section 6.3.2, testers had a difficult time
navigating within BGONE. Had this not been a research experiment, most agreed
they would have reverted to a text based data exploitation method to “get the job done
quickly”. The use and implementation of knowledge and techniques developed from
this study must therefore complement, not replace, current intelligence exploitation
methods. As with any updated technology or method of doing business, people need
time to conquer the learning curve before benefits can more accurately be assessed.
7.2.1 Implementation Consideration
The Project Broadsword program office, encouraged strongly by the user
community, is looking to expand into the three-dimensional realm when using data
visualization methods. Although they wish to give the users what they want, I feel
that jumping headfirst into such an undertaking is a mistake. At this time, the
60
maturity level of the technology and standard computer input tools supplied to each
analyst are not yet sufficient. Rather than a full implementation of three-dimensional
data visualization, as presented, a gradual integration of this technology should be
explored. One possible technique involves coupling results from the current two-
dimensional geospatial map with BGONE enabled exploitation. When an analyst is
presented with results, as seen in Figure 7.1, rather than the green icon leading to a
Figure 7.1: Current Results Display (Color Plate 20)
text-based format, a new window such as Figure 7.2 could be displayed. This new
window shows all products available at the chosen location of interest. With the
results map left in an unmodified window, the analysts can visually explore the
locality without losing global perspective. Applying the concept of Figure 7.2, the
new data navigational techniques allow for quick exploitation of products of interest
without physical movement within the three-dimensional space. This provides
improved intelligence exploitation techniques, as discussed in Section 7.2, while
alleviating the discomfort testers felt while traversing a VRML map. This
research set out to develop new data exploitation methods and techniques for the
plethora of digital intelligence products. Implementing the data navigation
techniques, as described in Section 5.3.2, required the creation of new methods that
allow automatic pop-up dialog box generation in a VRML environment. This
capability, never before seen in this format, is crucial for the effective use of data
61
ffVBrottdswnrd HP Navigational Capabilities Demo - Netscape
I Wdwiietu: : —f— ;- -—r—:- ,
-. J| % r Project Broadsword
Project Broadsword is an effort to provide an
extensible framework of services and tools which
help a user gather intelligence information from
distributed data sources in support for his/her
mission. Broadsword is divided into three
functional areas: Gatekeeper (server). User
Services (client), and Additional Services.
Broadsword can also be found on the following
networks:
Collateral: http://199.55.159.11:8080/bsword
Intclink: http://webl.rome.ic.gov/bsword
Paaemaster email
Technical POC email
Figure 7.2: BGONE Enhanced Results (Color Plate 21)
visualization when the amount of data to be exploited exceeds the handling
capabilities of the graphics. The application of these techniques can and should be
applied to other data visualization problems where a large amount of information is
present. These other problems include data exploration in a two-dimensional
environment, which highlight the fact that the creative handling of data in two-
dimensional visual space has not yet been exhausted.
7.3 Further Research
This thesis set out to demonstrate the benefits of exploiting intelligence
products using data visualization employed in three-dimensional space. However,
62
research revealed that more work is needed before successful implementation in this
area. Numerous areas of study need to be pursued, to include correlation of a
products’ representative size to the users’ view level, virtual reality device usage, and
enhanced multimedia capabilities.
Under the current architecture, proper scaling between a glyph or an icon with
respect to its location on the geospatial map is not performed. As a user moves about
the map or zooms in and out, the icon size does not change with respect to the users’
view. For example, at one level of resolution a single result’s icon can encompass
fifty square nautical miles of area, while in another the same results icon may cover
only one square mile. Although the glyph size does change with respect to a users’
view, the geospatial location over which it is located is not defined in exact detail
since this information is not passed to the graphics Tenderer. The design process
currently arbitrarily sets glyph and icon sizes. A method for properly controlling and
varying the size of the glyph or icon with respect to the actual data location as the
analyst changes worldview perspective needs to be addressed.
Since most analysts are constrained physically by two dimensional control
devices, effective ways of breaking out of this confinement need to be explored.
Using virtual reality devices, such as a space ball or head tracker, would allow
additional freedom and ease of movement. Inclusion of these devices may effectively
eliminate the movement difficulties outlined in Section 6.3.2.
This research has only brushed the surface of what can be achieved with
multimedia. As network bandwidth and computer processing increases, the
63
possibilities for video streaming (both live and archived) and other technologies hold
many promises for future enhancements.
64
Color Plates
QUERY:
PRD.PRODEMT =
"TTFF6.0”
Revise Query
Save Query As
Color Plate 2: Preattentive Processing
Color Plate 3: Symbolic Glyphs
66
Color Plate 4: Multilevel Glyphs
Color Plate 5: Cylinder Shaped Level One Glyphs
67
Color Plate 6: Cube Shaped Level One Glyphs
Color Plate 7: Transparency as Designed
68
Color Plate 8: Transparency in Operational Environment
Color Plate 9: Redesigned Level One Glyph
69
Color Plate 10: Abstract Level Two Glyph
Color Plate 11: Model Based Level Two Glyph
70
Color Plate 12: Low Level Of Detail Glyphs
Color Plate 13: Low Level Of Detail Glyphs on a VRML Landscape Map
71
Color Plate 14: Proximity Sensor Activation Rendering
Broadsword VRML Map - Netscape K3
- — - - —l
Color Plate 15: Products Represented by Gray Glyph
72
file Edit yiew Go Commumcatci yelp
■$ y 3 ?&
& 2d
Search Netscape
a
SscurUy
a
Shop
' Bookmorts X Loca>ion:|hley//q,'Michael G/ThesisMariel/BGONE hrm
.♦ Project Brwufaverd
Project Broadsword is an effort to provide an extensible
framework of services and tools which help a user gather
intelligence information from distributed data sources in support
for his/her mission. Broadsword is divided into three functional
areas: Gatekeeper (server), User Services (client), and
Additional Services.
Broadsword can also be found on the following networks:
Collateral: http://199.55.159.ll :8080/bsword
Int clink: http://webl.rome.ic.gov/bsword
Color Plate 16: VRML Environment with Frames
73
dt ’■**■ OocumentDcm*
Dialog box generated
by the isOver method
SI
3 _
F-1SE Strike Eagle
Mission
The F*15E Strike Eagle is a dual-role fighter designed to perform
air-to-air and air-to-ground missions. An array of avionics and
electronics systems gives die F-15E the capability to fight at low
altitude, day or night, and in inclement weather and perform its
primary function as an air-to-ground attack aircraft.
Features
The aircraft uses two crew members, a pilot and a weapon
systems officer. Previous models of the F-15 are assigned
air-to-air roles; the "E" model is a dual-role fighter. It has the
capability to fight its way to a target over long ranges, destroy
enemy ground positions and fight its way out
An inertial navigation system uses a laser gyro to continuously
monitor the aircraft's position and provide information to the
central computer and other systems, including a digital moving
map in both cockpits.
The APG-70 radar system allows air crews to detect ground ^
•<* •*-. l~S3 A
Data placed in frame via
the isPressed method.
Color Plate 17: Text Handling
74
Color Plate 18: Glyph Representation
Single Green Square
Color Plate 19: Two-dimensional Geospatial Map
Color Plate 20: Current Results Display
Project Broadsword is an effort to provide an
extensible framework of services and tools which
help a user gather intelligence information from
distributed data sources in support for his/her
mission. Broadsword is divided into three
functional areas: Gatekeeper (server), User
Sendees (client), and Additional Services.
in
Broadsword can also be found on the following
networks:
Collateral: http://199.55.159.11:8080/bsword
Int clink: http ://web 1. rome. ic. gov/bsword
Pacemaster email
Color Plate 21: BGONE Enhanced Results
76
Bibliography
1. Lee Paschall, “C3I and the National Military Command System”,
Command and Control Seminar, Harvard University, 1980
2. Memorandum to Dr. McCall from General Fogleman, CSAF and Dr.
Widnall, SecAF, entitled “New World Vistas Challenge for Scientific
Advisory Board (SAB)”, 29 November 1994
3. USAF Scientific Advisory Board, “New World Vistas Air and Space
Power for the 21 st Century,” Summary Volume, December 1995
4. William E. Colby, “The Developing Perspective of Intelligence”,
Command and Control Seminar, Harvard University, 1980
5. Christopher G. Healy, “Choosing Effective Colours for Data
Visualization”, Proceedings of IEEE Visualization '96, 1996
6. L. D. Bergman, B. E. Rogowitz, L. A. Treinish, “A Rule-Based Tool for
Assisting Colormap Selection”, Proceedings of Visualization ’95, 1995
7. B.E. Rogowitz, L .A. Treinish, “An Architecture for rule-based
visualization. In Proceedings Visualization ’93, 1993
8. John W. Senders, Marc Green “Expert Witness and Human Factors
Design”, http://www.ergogero.com, 1999
9. W. Cleveland, “The Elements of Graphing Data”, Wadsworth, Command
and Control Seminar, Harvard University, 1984
10. R. Abarbanel, “Problem Solving with and without Visualization”,
Proceedings of Visualization ’93, 1993
11. D. Woods, “The Cognitive Engineering of Problem Representations”, In
G. Weir and J. Alty, “Human-Computer Interaction and Complex
Systems”, Academic Press, 1991
12. Alan M. Davis, “Software Requirements”, University of Colorado at
Colorado Springs, Prentice Hall PTR, 1993
13. “ISO/IEC 14772-1:1997 Virtual Reality Modeling Language (VRML97)”,
htttp://www.web3d.org/Specifications/VRML97, 1997
77
Vita
Lieutenant Michael L. Goeringer was bom on 3 September 1967 in Springfield
Illinois. He graduated from La Cholla High School in Tucson, Arizona in June 1985. He
started his military career as an Airman Basic in January 1987. His first assignment was
at the 341 st Operational Missile Maintenance Squadron, Malmstrom AFB as a missile
maintenance technician for the Minuteman III weapon system. He entered undergraduate
studies at Park College through the on base education center, earning his Bachelor of
Science degree in Computer Science in 1995.
Lieutenant Goeringer earned his commission through Officer Training School in
May 1996. His first commissioned assignment took him to Rome Laboratory in Rome,
New York. He served as the program manager for the Global Awareness Virtual
TestBed in support of the Air Force Office of Scientific Research mission. In August
1998, he entered the Graduate School of Engineering and Management, Air Force
Institute of Technology. Upon graduation he will be assigned to the College of
Aerospace Doctrine, Research and Education, Maxwell AFB, Alabama.
78
REPORT DOCUMENTATION PAGE
Form Approved
OMB No. 0704-0188
___I_
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,
gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection
of information, including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports
(0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be
subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. _
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE
xx-03-2000 Master’s Thesis
3. DATES COVERED (From - To)
Feb 1999 - Mar 2000
4. TITLE AND SUBTITLE
THREE-DIMENSIONAL DATA VISUALIZATION OF ELECTRONIC
MILITARY INTELLIGENCE USING THE PROJECT BROADSWORD
SYSTEM
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S)
Goeringer, Michael L, Lieutenant, USAF
5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
Air Force Institute of Technology
Graduate School of Engineering and Management (AFIT/EN)
2950 P Street, Building 640
WPAFB, OH 45433-7765
8. PERFORMING ORGANIZATION
REPORT NUMBER
AFIT/GCS/ENG/OOM-08
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
Dr. John Salerno
32 Brooks Road
Rome, New York 13441
Phone: (315) 330-4429 DSN:587-4429
Fax: (315) 330-3913
10. SPONSOR/MONITOR’S ACRONYM(S)
AFRL/RRS IFED
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.
13. SUPPLEMENTARY NOTES
14. ABSTRACT
This thesis explores procedures for enhancing the capabilities of Project Broadsword, an intelligence data retrieval system,
using three-dimensional data visualization. Employing a stand-alone representative environment, this research develops methods
and techniques for overcoming problems encountered when visualizing large quantities of data. Textual handling through graphical
triggers is also addressed.
The results of this research demonstrate the effects of utilizing three-dimensional visual cues to organize, perceive, and navigate
vast amounts of intelligence products and their associated metadata. Conclusions drawn from this research directly affect the next
major release of the Project Broadsword system, currently under development.
15. SUBJECT TERMS
Three-dimensional Data Visualization, Intelligence Products, Project Broadsword, Computer Graphics
| 16. SECURITY CLASSIFICATION OF:
17. LIMITATION OF
ABSTRACT
18. NUMBER
OF
PAGES
91
19a. NAME OF RESPONSIBLE PERSON
Lt. Col. Timothy Jacobs, ENG
a. REPORT
b. ABSTRACT
c. THIS PAGE
u
u
u
UL
19b. TELEPHONE NUMBER (Include area code)
(937) 255-6565 x4279 / DSN 785-6565 x4279
Standard Form 298 (Rev. 8/98)
Prescribed by ANSI Std. Z39.18
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11