Skip to main content

Full text of "DTIC ADA380757: Three-Dimensional Data Visualization of Electronic Military Intelligence Using the Project Broadsword System"

See other formats



Michael L. Goeringer, Lieutenant, USAF 




Wright-Patterson Air Force Base, Ohio 


20000815 197 




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 


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. 



Michael L. Goeringer, B.S., 
Lieutenant, USAF 


’ a 7 SA'vr? 


Dr. John J. Salerno 

^cw OC 


I l/Udhl' 

Maj. I 

8 tfA&r gooo 



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 


Table of Contents 


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 


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 Level One Glyph.34 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 


7.2.1 Implementation Consideration.60 

7.3 Further Research. 62 

Color Plates. 65 

Bibliography. 77 

Vita. 78 


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 


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 




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 

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. 




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. 


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 


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 


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. 


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. 


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. 


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 


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). 


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 





Revise Query 
Save Query As 


[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 




Revise Query 

y ujdivi . _ 


"T1EE6.0" Save Query As 

Source _I_ Query Status 


10 Hits 


| 1998 [ 



[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. 


■“■“^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 





Zoom: | AUTO "3 [Current: Years] 

J19 9 9 0 3 3119 4 8 5 9 

Update Display | 

ypertext Link 

Timeline Heading, 

Figure 2.2: Timeline Results 





Revise Query 

Save Query As 

Green Two-dimensional Icon 

Figure 2.3: Geospatial Map Results (Color Plate 1) 


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 


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 


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 


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 


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 


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 


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. 


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 


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 

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. 


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 


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 


Task #2: Retrieve all imagery products that have IPL as 

a data source. 

Task #3: Identify all F-15E aircraft imagery 


Intelligence Product-to-Glyph Color Identification Mapping 


Product Type 










Multiple Products 

Figure 3.1: Experiment Scenario 


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 


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. 


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 


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 


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 


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. 


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 


further discussed throughout this chapter. 

Table 5.1: Intelligence Data Mapping 

Product Type 

Visual Attribute - Color 









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 

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 


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) 


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 


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) 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 


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) 


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 


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. 


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. 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 


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 


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) 


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 


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. 


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 


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 


Allows the user to move through the scene 
in an X and Z direction 


Allows the user to move through the scene 
in an X and Y direction 


Allows an entire scene to be rotated around 
its own X axis 


Allows an entire scene to be rotated around 
its own X and Z axis 


Allows the user to move through the scene 
in an X, Y, and Z direction. 


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 


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 

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 


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. 


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. 


|Pnr* S*curty Shop 


3 _ 

Dialog box generated 
by the isOver method 

F-15E Strike Eagle 


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 


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) 


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 

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 

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. 


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. 


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 

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 


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 


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- 


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. 


ttf "" Document Done 

gStartj JOemc3 j Jhogeaot* j JKfahadG j JkjGAggjicJ .jTtesis Model] jAugns | SOtoeen ca... \ jjMyCompi*;t| _j(Cj 

F-15E Strike Eagle 


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 


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 


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. 


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 


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 

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 


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 


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 


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, 


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 


possibilities for video streaming (both live and archived) and other technologies hold 
many promises for future enhancements. 


Color Plates 



Revise Query 

Save Query As 

Color Plate 2: Preattentive Processing 

Color Plate 3: Symbolic Glyphs 


Color Plate 4: Multilevel Glyphs 

Color Plate 5: Cylinder Shaped Level One Glyphs 


Color Plate 6: Cube Shaped Level One Glyphs 

Color Plate 7: Transparency as Designed 


Color Plate 8: Transparency in Operational Environment 

Color Plate 9: Redesigned Level One Glyph 


Color Plate 10: Abstract Level Two Glyph 

Color Plate 11: Model Based Level Two Glyph 


Color Plate 12: Low Level Of Detail Glyphs 

Color Plate 13: Low Level Of Detail Glyphs on a VRML Landscape Map 


Color Plate 14: Proximity Sensor Activation Rendering 

Broadsword VRML Map - Netscape K3 

- — - - —l 

Color Plate 15: Products Represented by Gray Glyph 


file Edit yiew Go Commumcatci yelp 

■$ y 3 ?& 

& 2d 

Search Netscape 





' 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: 

Color Plate 16: VRML Environment with Frames 


dt ’■**■ OocumentDcm* 

Dialog box generated 
by the isOver method 


3 _ 

F-1SE Strike Eagle 


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. 


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 


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. 


Broadsword can also be found on the following 

Int clink: http ://web 1. rome. ic. gov/bsword 

Pacemaster email 

Color Plate 21: BGONE Enhanced Results 



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”,, 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://, 1997 



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. 



Form Approved 
OMB No. 0704-0188 


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. 



xx-03-2000 Master’s Thesis 

3. DATES COVERED (From - To) 

Feb 1999 - Mar 2000 







Goeringer, Michael L, Lieutenant, USAF 





Air Force Institute of Technology 

Graduate School of Engineering and Management (AFIT/EN) 

2950 P Street, Building 640 

WPAFB, OH 45433-7765 





Dr. John Salerno 

32 Brooks Road 

Rome, New York 13441 

Phone: (315) 330-4429 DSN:587-4429 

Fax: (315) 330-3913 








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. 


Three-dimensional Data Visualization, Intelligence Products, Project Broadsword, Computer Graphics 







Lt. Col. Timothy Jacobs, ENG 








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