<%BANNER%>

Architecting Rube Worlds: a methodology for creating virtual analog devices as metaphorical representations of formal systems

University of Florida Institutional Repository
Permanent Link: http://ufdc.ufl.edu/UFE0000670/00001

Material Information

Title: Architecting Rube Worlds: a methodology for creating virtual analog devices as metaphorical representations of formal systems
Physical Description: Mixed Material
Creator: Damkjer, Kristian Linn ( Author, Primary )
Publication Date: 2003
Copyright Date: 2003

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000670:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000670/00001

Material Information

Title: Architecting Rube Worlds: a methodology for creating virtual analog devices as metaphorical representations of formal systems
Physical Description: Mixed Material
Creator: Damkjer, Kristian Linn ( Author, Primary )
Publication Date: 2003
Copyright Date: 2003

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000670:00001


This item has the following downloads:


Full Text











ARCHITECTING RUBE WORLDS:
A METHODOLOGY FOR CREATING VIRTUAL ANALOG DEVICES AS
METAPHORICAL REPRESENTATIONS OF FORMAL SYSTEMS
















By

KRISTIAN LINN DAMKJER


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003
































Copyright 2003

by

Kristian Linn Damkjer
































Dedicated to my wife Bryony,
my parents Keith and Deborah Damkjer,
and my sister April Derato,
for their constant support and encouragement















ACKNOWLEDGMENTS

I would like to sincerely thank my advisor and committee chairman, Dr. Paul A.

Fishwick, for his constant instruction, guidance, and support throughout my graduate

studies at the University of Florida. I would also like to extend my gratitude to Dr.

Abdelsalam Ali Helal and Dr. Joachim Hammer for their time and interest in serving on

my thesis committee.

I sincerely appreciate the time and devotion of my companions on the University of

Florida RUBE Project, Mr. Minho Park, Mr. Jinho Lee, and Ms. Hyunju Shim, who made

this research truly an enjoyable experience.

Many thanks go to my fellow graduate students in the digital arts and sciences

program, especially Mr. John Hays and Mrs. Joella Walz, for their companionship and

support through the highs and lows of graduate studies. I will never forget their

friendship.

I am deeply grateful to my parents, Keith and Deborah Damkjer, for their constant

prayer, support and encouragement, and to my wife, Bryony, whose love, support, and

willingness to share these last two wonderful, though often trying, years with me will be

a treasured memory for years to come.

Finally, I owe an extreme debt of gratitude to everyone who has prayed for me

throughout my academic endeavors and to God for His many blessings; may He continue

to guide me and my family throughout our lives.
















TABLE OF CONTENTS
Page

A C K N O W L E D G M E N T S ................................................................................................. iv

L IST O F T A B L E S ......................................................................... ..... ....... ix

LIST OF FIGURES ............................... ... ...... ... ................. .x

A B S T R A C T .......................................... .................................................. x iii

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

A Brief H history of RUBE....................................................... .... ...................... 1
C h a lle n g e s........................................................ ................ .. 3
Organization of Thesis ............... .................................................4

2 B A C K G R O U N D ............................................................ ........ .......... .... ....

M them atics as a Language ............................................................ ............5
M them atics in M odel D esign and Sim ulation ........................................ .................6
S c h e m a s ............................................................. ................ 7
Causation and A nim ism .................... .. .. .... .................. .................................. 10
Manipulative Objects: Play and the Emotional Factor...............................................11
Im m version and Engagem ent ......................................................... .............. 12
Multi-Sensory Model Representation..................... .............................. 13
S u m m ary ...................................... .................................................. 13

3 CONCRETIZING CONTENT DEFINITION ................................. ...............15

O v e rv iew ......................................................... ............... ................ 1 5
R efining the M X L Schem a ........................................................................... .... 15
M X L D ata Types .................. ..................................... ...... .......... 16
MXL Root and Top-Level Elements......................................................17
The M XL root elem ent............................... ......... .................. 17
The m odels group ................................................ .. .... .. .......... 18
The sim ulation elem ent ......... ................. ........................... ............... 18
Removal of the comment element ............................................19
M odel Definition Elements and Groupings.............................. ............... 19
H andles: input and output ........................................ ........................ 20



v









M odel categories .......................... ....... .. ............. ...............21
Generalizing Node Definition Structure..................... ........... ............... 21

4 EXTENDING RUBE TO SUPPORT THE MULTI-MODELING PARADIGM...... 24

M X L External M odel File R eferencing........................................... .....................24
M X L M massive M erge......................................................................... ...................26
M odel L ib raries ................................................................2 7

5 M O D EL D EFIN ITION ........................................... .................. ............... 28

O overview .................................. ................. ...... ................... 28
Identification of Model Subcomponents ....................... .....................28
Self-Similarity of Model Functionality and Interface .............. ...................28
Logical and/or Functional Groupings............................... ....... ............... 29
O bject-O oriented A approach ............................................................. .................. 30
Optim izing M odel D definition ............................................................................. 30

6 METAPHORS TO MODEL BY ...................................................... ...................32

M odel Structu re ................................................................33
M odel M etaphors.......... .............................................................. ... .... ....... 33
M odels are environm ents ........................................ ......... ............... 33
M odels are containers........................................................ ............... 33
M odels are buildings ........................... ............... ...... .. ........ ... 34
M odels are m machines ......................................................... ............... 34
H andle M etaphors .................................. .. .. ...... ............ 34
H andles are thresholds ........................................ ........................... 34
H andles are valves .............................. ........... ........................35
Handles are ports or outlets ................................ ... .............................. 35
H andles are radio tow ers ............................................................ ........ 35
H andles are puzzle pieces ........................................ ........................ 35
H andles are fasteners......................................................... ............... 36
N ode M etaphors ........................................ .... ....... .... ....... 36
N odes are containers ......................... ...... .............. .......... .. ................. 36
N odes are room s ................... .................... .. ... ...................... 37
N odes are m machine parts ........................................ ......................... 37
E dge M etaphors ..................................................... ... .. ........ .... 37
Edges are conduits .............................. ........... ........................38
Edges are a line-of-sight..................... ..... ........................... 38
M o d el E x ecu tio n ................................................................................................... 3 8
Input M etaphors ..................................38.............................
Input is fuel ............. ................... ......... ............. ..... .. ... 39
In p u t is a so u rc e ............ ......................................................... .... .... .. .... 3 9
Input is an agent entering ........................................ ......... ............... 39
Input is a m essag e ................................ .............. .................. ..............39
Input is an action ............................ ................................ .... ..... ...... 40









D ata Flow M etaphors ................................................. ............................. 40
D ata flow is a transfer of energy ...................................... ............... 40
D ata flow is flow ing w ater................................ ......................... ....... 40
D ata flow is a m oving agent................................... ......... ............... 41
D ata flow is a transm mission ........................................ ....... ............... 41
A ctive N ode M etaphors......................................................... .............. 41
A ctive node is anim ation......................................... ......................... 41
A ctive state is agent location.................................... ........................ 42
A ctive state is a color or texture................................................................ 42
A ctive node is a sound ........................................ ........................... 42
O utput M etaphors .................... ........ ............................................ .. ..............42
Output is a manufactured product .... .......... .......................................42
O utput is a sound .................. .................. ...... .................... .. 43
O utput is an agent exiting....................................... .......................... 43
O utput is a reply ..................... .. .. ..................... .... .. ........... 43
O utput is a reaction ............................................. .................. ..............4 3
O utput is a piece of art ........................................ ........................... 43

7 MODEL PRESENTATION DESIGN METHODOLOGY.....................................44

M odel P rototypes............... .. ........ .......... ...... ....................... 44
Using Groups and Transforms to Mimic Content Structure.................. ............46
L logical G rou p s ..............................................................4 6
C o n tain ers ................................................................................ 4 7
M odel H ierarchy and G eom etry ........................................ ........................... 49

8 C O M M E N T P A R SE R ........................................................................ ..................52

D description ...................................................................................................... ....... 52
S tru c tu re .........................................................................................5 2
M a p p in g .........................................................................................5 6

9 BA RN SLEY FERN IFS ...........................................................63

D description and D definition ................................................................................... 63
S tru c tu re ........................................................................................6 6
IFS FBM .......................... ........ ............................... ......... 66
Transform action Selection FSM ............................................... ................ 67
Augmented Affine Transformation Matrix Calculation FBM .........................68
Sine, Cosine, and Negative Sine Calculation FBM.......................... 69
Three-By-Three By Three-By-Three Matrix Multiplication FBM ..................69
Row-Column Product (Vector Multiplication) FBM ..........................................70
N ew Point Calculation FBM ......................................................................... 71
M ap p in g ..........................................................................................................72









10 C O N C L U SIO N ......... ......................................................................... ........ .. ..... .. 75

Summary of Results and Future W ork .................................................................... 75
A uxiliary O bjectives.......... ................................................................ .... .... ..... 76
Autom action ................................... .................. ........... .........76
Extension to Representations other than VRML.............................................77

APPENDIX

A M XL SCH EM A D EFIN ITION ......................................................... .....................78

B MODEL PRESENTATION EXAMPLE STRUCTURE ..........................................85

C MODEL DEFINITION FOR THE C-STYLE COMMENT SCANNER ..................89

Com m ent Scanner FBM : com m ent.xm l ........................................ .....................89
Comment Scanner FSM: comment sub.xml.......................... .... ........... 90
Comment Scanner JavaScript: comment.j s ..................................... ........... ..........91

D MODEL DEFINITION FOR THE BARNSLEY FERN IFS................................97

B am sley F ern IF S: B am sley.xm l .................................................... .....................97
IFS States: States.xm l ................................................. ........... ........ .............. .. 100
Affine Transformation Matrix Calculation: AffineCalc.xml............................... 102
Sine and Cosine Calculation: SinCosCalc.xml..................... ................... 108
3-by-3 by 3-by-3 Matrix Multiplication: MatMult.xml ................. ... .................109
Row-Column Vector Multiplication: RCCalc.xml.................................................114
IFS Attractor Point Calculation: PointCalc.xml ...................................................115
IFS-Specific Semantics JavaScript: Barnsley.js ................................................... 117
RUBE Math Library Functions JavaScript: math.js.................. ......................... 121

L IST O F R E FE R E N C E S ........................................................................ ................... 122

BIOGRAPHICAL SKETCH ............................................................. ............... 124















LIST OF TABLES

Table p

3-1. List of original MXL elements with descriptions....................................16

3-2. M XL data types ..................................... .................... ........... ....... ..... 17

3-3. M X L m odel elem ents ....................................................................... ...................22

8-1. Scanner transitions summarized ................................................... ...................55

9-1. A fine transform nations in R 3 ............................................ ............................. 65















LIST OF FIGURES

Fiure p

1-1. RUBE architecture ....................... .... .............. .............................. .2

3-1. M X L top-level elem ents.................................................. .............................. 18

3-2. The m models group .......................................... ............. .... ....... 20

3-3. The handles group .................................... ..... .......... .............. .. 20

4-1. Parent m odel external reference ........................................ .......................... 25

4-2. Externally defined sub-model (mySubModel.xml)................................................25

7-1. Sim ple FB M block diagram ........................................................... .....................44

7-2. Sim ple FB M M X L definition ........................................................................ .. .... 45

7-3. A VRML presentation file showing model prototypes ................... .............. 46

7-4. Model component grouping in presentation files......................... .. ...............47

7-5. C containing transform s ....................................................................... ..................48

7-6. Prototype instances and atom ic structures.............................................................. 50

8-1. C om m ent scanner F SM ........................................... ....................................... 53

8-2. C om m ent scanner FB M .................................................. ............................... 53

8-3. Com m ent scanner virtual w orld ........................................ .......................... 56

8 -4 B lo ck : F ile ......................................................................... 5 8

8-5 B lo ck : S can n er.................................................. ................ 5 8

8 -6 B lo c k : P a rse r........................................................................................................ 5 9

8 -7 S tate : P ro g ra m ..................................................................................................... 5 9

8 -8 S tate : S la sh .................................................................6 0









8 -9 S tate : C P P ......................................................................................................6 0

8 -10 S tate : C ................................................................6 1

8 -1 1. S tate : S tar.......................................................6 1

8 12 T ra n sitio n s ............................................................................................................... 6 2

8-13. Signage indicating transition directionality ......................................................62

9-1. The general IFS FB M ..................................................................... 66

9-2. T ransform ation states F SM ................................................................................. 67

9-3. Augmented affine transformation matrix calculation................ ............... 68

9-4. Sine, cosine, and negative sine calculation FBM ..............................................69

9-5. M atrix m ultiplication F B M ................................................................................. 70

9-6. Row-column product calculation FBM ............. ............................ 71

9-7. New point calculation FBM .............................................. ...............72

9-8. B arnsley C observatory ............................................................72

A-1. MXL schema definition........................................78

B-1. Model presentation example file.................................... ...... .........85

C -1. com m ent.xm l ................................................................89

C -2 com m ent su b .x m l ............................................................................................... 9 0

C -3 c o m m e n t.j s ................................ ......................................................................... 9 1

D -1. B arn sley .xm l ....................................................... 97

D -2 States.xm l .................................................................................... 100

D -3 A ffin e .x m l ............. .................. ................................................................ 10 2

D -4. SinC osC alc.xm l ....................................................... 108

D -5 M atM u lt.x m l .................................................................................................. 10 9

D -6 R C C a lc .x m l .................................................................. ..................................1 14

D-7. PointCalc.xml ................................................. ..................... 115









D -8. B am sley.js .............................................. 117

D -9 m ath .j s....................................................12 1















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

ARCHITECTING RUBE WORLDS:
A METHODOLOGY FOR CREATING VIRTUAL ANALOG DEVICES AS
METAPHORICAL REPRESENTATIONS OF FORMAL SYSTEMS

By

Kristian Linn Damkjer

May, 2003

Chair: Paul A. Fishwick
Major Department: Computer and Information Science and Engineering

Several modeling and simulation toolkits have been created to help facilitate the

model development process and aid in extending models and simulations into new

arenas. New ways of visualizing models and simulations have likewise been developed

to take advantage of technologies available in these new arenas. RUBE is a project that

uses a combination of XML technologies, Java, JavaScript, and the Virtual Reality

Modeling Language (VRML), to provide a framework for developing dynamic models

that run as virtual environments on the Internet. In RUBE, model structure is independent

of its presentation, which allows for improved sharing of models and also for presentation

customization.

My research attempts to establish a methodology for creating presentation worlds

that ensures proper correlations between the model content and presentation. My

approach uses virtual analog devices and architectural constructs as metaphorical

representations of model structure, though I also provide a generalized methodology for









using metaphorical mappings and grouping structures to ensure strong correlations. This

research has also resulted in the restructuring of the Multi-model Exchange Language

(MXL)-one of several XML applications employed in RUBE-for better model content

definition, as well as extended support for multi-modeling in RUBE.















CHAPTER 1
INTRODUCTION

A Brief History of RUBE

The current modeling and simulation endeavor entitled RUBE originates from prior

research and development that started in the early 1990s with the creation of code

libraries-consisting of C source, libraries and executables-entitled SimPack (1990).

Significant portions of SimPack were dedicated to handling future event lists and queuing

(both of which are bases for discrete event simulation, a popular method for simulating

formally defined dynamic systems) [10]. These portions of SimPack underwent several

revisions resulting in OOSIM (1995), coded in C++ using the object-oriented paradigm

[10], and SimPack J/S (2002), a JavaScript and Java port developed with the intent of

supporting web-based simulation and which now serves as the core simulation engine for

RUBE [16].

As the names suggest, the focus of SimPack and its successors is model simulation,

not model development. To augment the simulation toolkits, specifically SimPack and

OOSIM, with multi-model development capabilities, the Object-Oriented Physical Multi-

modeling (OOPM) toolkit (1996) was developed [9]. This toolkit provided a graphical

interface for developing multi-models with support for several model types and is

regarded as the predecessor to the RUBE Project [14].

The RUBE Project began in 2000 as a continuation of research in multi-modeling

and web-based modeling and simulation with the initial purpose of facilitating dynamic

multi-model construction and component reuse in three-dimensional distributed












immersive environments. To this end the Virtual Reality Modeling Language (VRML)


was employed as the primary language for specifying scene geometry and presentation


dynamics. Initial work in this arena was done using the VRML "Prototype" extensibility


mechanism [10]; however, that approach has recently been replaced with the Extensible


Markup Language (XML) based framework that now serves as the basis for RUBE [14].


This approach more strongly decouples the model content from the model presentation


which in turn allows for better component reuse and customization of model


representation.


The initial framework designed by Kim has been modified slightly to accommodate


recent developments in our research and the current structure is outlined in Figure 1-1.


Scene File (VRML)


MXL to DXL Translator


NISTVRMLtoX3D DXL


DXL to J/S Translator






Model Fusion Engine (XSLT)


X3D


X3D to VRML (XSLT)


Simulation File (VRML)


3 User Generated Files
E rube Generated Files
0 rube Generated Files Requiring User Edits
E Translation Engines


Figure 1-1. RUBE architecture









In the RUBE pipeline, the developer provides as input scene files describing how the

model should be presented, model files defining the model topology, and semantic files

defining how the model should execute (simulation rules). These files are then merged

using XML Style-sheet Language Transformation (XSLT), Document Object Model

(DOM), and custom-developed translation mechanisms to produce a single presentation

file with an interface for initializing simulation execution and hooks for simulation

output. The developer modifies this file to use the simulation output data to drive the

presentation dynamics while the simulation executes.

Challenges

With more robust modeling and simulation toolkits-affording greater possibilities

in model definition, presentation, simulation, and distribution-and the fact that model

definition is ideally decoupled from its presentation, the need for a formalized

methodology for developing models and their corresponding presentations increases.

To this end the primary objectives of this research are to refine the Multi-model

Exchange Language (MXL) to remove ambiguities in model definitions and to facilitate

language extensions in the form of adding new model types; to extend the RUBE

framework to allow for model definition and semantic component reuse; to explore

model definition development to take advantage of the multi-model paradigm; to explore

metaphorical mappings for model representations; and to establish a methodology for

creating presentation worlds for models developed with the RUBE framework. Auxiliary

objectives of this research are to explore the possibility of partially automating the

established methodology and to investigate extending the methodology to presentation

vehicles other than VRML.









Organization of Thesis

A philosophical background of the RUBE approach to modeling and simulation is

discussed in Chapter 2. Chapter 3 details the revisions made to MXL and the model

semantics definition structure. Chapter 4 details the extensions to the RUBE framework to

fully support multi-modeling. Chapter 5 explores model definition strategies. Chapter 6

outlines metaphorical mappings to use in model representations. Chapter 7 discusses the

methodology to be used in presentation development. Chapters 8 and 9 are summaries of

case-study implementations. Conclusions, future work, and auxiliary objectives are

discussed in Chapter 10.















CHAPTER 2
BACKGROUND

Mathematics as a Language

Mathematics is often thought of as an eloquent language, highly formalized and

structured, able to capture endless representations of phenomena in the world around us

succinctly and efficiently. It has its own syntax, grammar, and an alphabet that is the

product of centuries of abstraction and stylization due to economics and efficiency [17].

As higher mathematics has evolved, new symbols, syntax, and grammar have been

created to extend the mathematical language, albeit non-uniformly and non-universally,

resulting in "dialects" of mathematics. In general, however, mathematics is still regarded

as a fairly universal language, allowing for the exchange of formalized ideas and

concepts in spite of socio-economic as well as natural language barriers.

The power of mathematics to represent formal structures is inherent in its abstract

nature. However, with this power of generalization comes a loss of inherent

comprehension. Gennady Uzilevsky's hierarchy of languages, which organizes

languages and codes based on their inherent comprehensibility among humans, places

mathematics into categories that are among the least inherently comprehensible: iconic

languages, symbols, and meta-languages. Hughes observes the following when

commenting on this hierarchy:

Only the top five languages are what we think of as "languages": these are the least
powerful of the 12 .... Most human communication is done by the lower seven,
of which the most easily named (in decreasing order of power) are gesture, images,
movements, emotional conditions, music and color. [13:180]









Thus, mathematics should be augmented with those languages that are more universally

comprehensible, a feat we accomplish or observe readily in literature, oration, and fine

art.

Mathematics in Model Design and Simulation

To model is to abstract from reality a description of a dynamic system. Modeling
serves as a language for describing systems at some level of abstraction or,
additionally, at multiple levels of abstraction .... Models are used for the purpose
of communicating with each other-the alternative being that two people wishing
to discuss dynamics would be forced to work with the real system under
investigation. [8:27]

Mathematics, because of its powerful ability to generalize formal structures, is a

natural choice for the language of modeling. It is very simple, in mathematical terms, to

describe well defined dynamic systems as sets of quantitative, stochastic, and logical

relations: The resulting deterministic system is defined as the following 7-tuple [8:46-

47]:

(T, U, Y, Q, 0,, A)

* T: the time set. For continuous systems T = R (the real numbers), and for discrete
systems T = Z (the integers).

* U: the input set. Contains the possible values of the input to the system.

* Y: the output set.

* Q: the state set.

* 0 : the set of admissible (or acceptable) input functions. T -> U

* : the transition function. : QxTxTxU -Q

* A : the output function. A : Q Y

Through mathematics we are able to not only formalize the structure of dynamic

systems, but also to virtually execute and analyze the systems. We can easily and safely









manipulate aspects of the system to observe how the system responds and behaves as a

whole. We can apply these observations back to the real-world phenomena being

modeled, provided that the model is accurate.

Despite this power, the primary purpose for modeling is not simply for description

and analysis of a system, but for communication with others. Unfortunately, the

traditional way of representing mathematical concepts with calligraphic notation often

fails in this endeavor [11]. We should attempt then to extend our representation of

mathematical models. After all, the exchange of ideas is the raison d'etre of modeling.

This extension of the mathematical model is the basis for the University of Florida's

RUBE Project [5, 8, 12]. Mathematics, in the traditional sense, has been used to represent

both the content and presentation of model information; RUBE [5, 12] attempts to separate

one from the other and allow users to interpret models using their own presentation

schemas [10].

Schemas

RUBE [5, 12] is perhaps unique in its approach to modeling and simulation in that it

takes a narrative approach to representing models. By doing so it attempts to draw on

schemas not only to improve comprehension of models, but also to create a synergistic

representation that addresses both of Bruner's "two modes of thought".

There are two modes of cognitive functioning .. each producing distinctive ways
of ordering experience, of constructing reality. The two (though complementary)
are irreducible to one another. Efforts to reduce one mode to the other or to ignore
one at the expense of the other inevitably fail to capture the rich diversity of
thought. [2:11]

The two modes of cognitive functioning are the paradigmatic, which attempts to achieve

an ideal formalization or mathematical explanation through categorization and

conceptualization of a system, and the narrative, which instead deals with the "human or









human-like intention and action and the vicissitudes and consequences that mark their

course." [2:13]

By incorporating schema and narrative in our model representation we are

attempting to improve comprehension of a model by providing an immediately

recognizable and familiar framework for the user. To successfully incorporate schemas,

however, we must first make some observations about how they are defined and operate,

then establish a means of merging the narrative structures with the paradigmatic ones.

There are several parallels between Bruner's schema definition and our formalized

model definition. Schemas are composed of states, crises, and redresses with the

possibility for cycles and branches in the narrative structure. The result of realizing this

structure is story [2]. Similarly, recall that dynamic systems are composed of states,

transitions, and next-states, and the result of simulating the system is output. The

narrative steady state is breached by some external force thus instigating the steady-state-

crisis-redress cycle [2]. In much the same way, systems transfer between states based on

input. Finally, both establish a chronological framework (even if narrative is not

presented in a strictly linear fashion). The mapping between the two is fairly direct:

schema-states map to system-states, crises map to transitions, redresses map to next-

states, steady-state breach forces map to system input, and story maps to system output.

Note also that we can extend this kind of schema mapping to system multi-

modeling in which we can have any combination of multiplicity, heterogeneity, and

hierarchy of models. Multiplicity implies several model instances, heterogeneity implies

several model types, and hierarchy implies several model layers in a nested structure

[10]. Black and Bower's observations on Episodes as Chunks in Narrative Memory









conform nicely to the multi-model structure. They observe that stories consist of

narrative elements that are interconnected to produce a coherent whole. They also

observe that the resulting network of interconnected elements can be represented in one

of two ways, either as consisting of several clustered sub-groups, or a resulting "macro"-

structure and "micro"-structure [1]. Both views parallel the multi-model paradigm and I

would further argue that neither is explicitly the normative structure.

To combine the paradigmatic and narrative modes of model representation in RUBE

we employ yet another narrative device: metaphor. We could have similarly chosen to

use the mathematical device of function. Both provide a direct mapping from one

domain to another; however, metaphor is explicitly a one-to-one mapping, a trait that

ensures consistency between the narrative schema and the formal system definition. Note

that this merge will result in not only a representation of the model output, but also of the

model structure and functionality. This representation goes far beyond that of the simple

graph or diagrammatic presentation [11]. Once the narrative world and the paradigmatic

model have been merged we can proceed to simulate the model. Assuming that the

model requires no direct input from the user to execute, what the user will observe are

elements of the narrative domain interacting with each other as a representation of the

model execution. Users will likely unconsciously compose a story based on the narrative

schema to describe and explain the interactions that they observe. This is a result of

causation and perceived intention as well as drawing on relevant scripts within the

schema to shape our perception, navigation, and interaction within a given scenario [6,

13].









Causation and Animism

Bruner recounts a discourse by Baron Michotte in which he demonstrates that when

subjects observed objects moving with respect to each other within highly limited

confines, the subjects impose causality. They saw the objects' movements as affecting

and being caused by the movements of other objects in the scene. It was found that the

spatial-temporal relationships of the objects could be manipulated to invoke various types

of behavior interpretation such as "launching," "dragging," and "deflecting" [2]. To

visualize this behavior, consider the classic arcade games of Pong and Arkanoid in which

a ball is "launched" across the screen at a target and is repeatedly "deflected" by other

objects in the scene.

Bruner further relates the work of Heider and Simmel who used methods similar to

Michotte's to demonstrate the irresistibility of "perceived intention". Subjects again

observed a short animated film and again bestowed behavior to the objects, yet this time

the behaviors were perceived as being intention-driven. Stewart demonstrated that again,

by manipulating the spatial-temporal relationships of the objects, other apparent

intention-driven behaviors could be invoked, for example: "searching," "pursuit," and

"persistence" [2].

Why are these results so interesting to RUBE model applications? The answer is

simple, they imply that users will create their own narrative to explain the model

presentation in terms that they already understand and will thus be able to achieve an

initial understanding of the model concept or perhaps a deeper comprehension. This

understanding will be made further concrete if the user already has a deep understanding

of the schema in which the model is being presented.









Recall, however, that one of the greatest strengths of the mathematical model is that

it allows us to safely and efficiently manipulate aspects of the system to observe how the

system as a whole responds [8]. To achieve this we must have some form of interaction

with the schema world and this is achieved through manipulative objects.

Manipulative Objects: Play and the Emotional Factor

By extending our representation of models to include schemas, we have extended

the ways in which users can understand our model. By introducing manipulative objects

in our schemas, we can allow the user to interact with and affect model execution. The

use of manipulative objects as a learning aid has been shown to be a positive influence on

learning and comprehension of mathematical concepts in general. This benefit has been

shown to hold across grade level, ability level, and topic. The use of manipulative

objects alone, however, does not guarantee successful comprehension of concepts [4]. In

RUBE, manipulative objects are incorporated in such a way that allows the user to play

within the confines of the model world in much the same way that a child might role-play

with toys [8]. This act of playing can contribute immensely to our comprehension of

systems and their complexity. Through this kind of hands-on exploration of "possibility

spaces" we glean terrific amounts of information about a system [13].

By incorporating interaction, the user also becomes an integral part of the narrative

schema and as a result tends to respond emotionally to events in the model world. Being

tightly coupled with the model world and having a high level of perceived control over its

outcome is amazingly parallel to computer gaming, which simultaneously engages

emotions of anticipation, joy, and acceptance. Should the model produce unexpected

results or behave in an unexpected manner (or even crash) even more emotions are

engaged resulting in surprise, anger, and frustration [13]. Coupling these emotional









responses with the presentation of the information has the added benefit of improving

comprehension and retention of concepts [3].

Hughes observes that, "Play, in the pure sense of simply controlling and

manipulating things, in a safe environment, for no particular purpose, is crucially

important for humans."[13:170] He notes that it is often easier to give users this "fun of

control" if there is an underlying serious didactic purpose and is surprised to find that

educational software is amazingly void of such "virtual toys" that allow for this kind of

playfulness and emotional reward [13]. Thus the potential for a system like RUBE is great

since it engages the user at so many levels and can also create an immersive experience.

Immersion and Engagement

The great potential of a system like RUBE for representing mathematical models is

that-beyond merely representing systems for communication and analytic purposes-

there exists potential for the experience of working with the models to be quite

pleasurable. Because RUBE is designed to operate in much the same way as interactive

narratives and computer games, this pleasure can be both immersive and engaging. The

immersive nature of RUBE is derived from the way that any given model presentation is

derived with a single schema in mind. While interacting with the single model

presentation the schema remains constant allowing the user to draw on the scripts

commonly associated with that schema to drive perception of the model and the various

ways to interact with it [6]. Of course, much of this is dependent on the model

presentation design. Interaction with the model should be well guided and should

conform to the schema world. Failure to do so will likely have the same result as in other

media: frustration [6]. RUBE invites users to have several immersive encounters with the

models by decoupling model content and presentation. This allows the user to apply









several presentations to the same model, much like a sort of "schema style-sheet". It is

also hoped that as users gain understanding of models they will seek out secondary

sources to augment their schemas for understanding the models and will then revisit the

RUBE models to gain greater insight through engaging experiences.

Multi-Sensory Model Representation

Finally, to enhance the immersive experience and provide further language cues,

the worlds that are developed for RUBE are usually multimedia applications. Humans are

quite obviously multimodal beings. Our senses are the mechanisms through which we

perceive our world and are integral to our understanding of it. Mixing several types of

media has the potential to be "sensory dynamite" [11]. Adding one form of media to

another can dramatically change our perception of the original media. Hollywood has

used this for years in movies. By laying a soundtrack under the visuals, directors are able

to subtly manipulate the mood of a scene and drive audience expectations. Incorporating

audio with graphics not only improves the perceived quality of the graphics, but also

provides a means of implying information about the graphics and their interactions [11].

In short, the more cues our model representation can give, the more modes of

interpretation we can potentially employ; however, Hughes cautions that the use of non-

verbal cues should be used subtly. We are highly sensitive to them and misuse

(intentional or not) results in being highly intrusive and can very easily have the opposite

of the desired effect [11].

Summary

We understand that traditional methods of conceptualizing and representing

dynamic systems models abstract the systems to such a level that they are no longer

readily comprehendible without formal training in simulation and model design. While






14


this level of abstraction and formalism is useful for the conceptualization and simulation

of the model, we attempt to counter the effects of abstraction on the presentation of the

model by using multi-media applications and schema frameworks in our worlds. This

approach has a high potential for creating immersive and emotionally engaging

experiences when working with the models. This in turn has benefits of improved

comprehension and retention of the information presented in the model.















CHAPTER 3
CONCRETIZING CONTENT DEFINITION

Overview

MXL was developed to allow for the formal definition of model topologies and the

establishment of simulation frames. The initial version of MXL displayed several

weaknesses. Prominent among these were tendencies of allowing ambiguity in model

definition and of being unintuitive due to its abstract nature as well as not truly allowing

component reuse. These weaknesses necessitated a restructuring of MXL. Since MXL is

an XML application, its syntax and grammar are defined using the XML Schema

Definition Language (XSD)-also an XML application. Thus, a restructuring of MXL

meant refining the MXL schema definition.

Refining the MXL Schema

The current version of MXL was created to address the following issues that were

surfaced in the original version:

* Model content and presentation were too tightly coupled.

* Model component granularity was too abstract.

* Subtleties of unique model types were impossible to implement.

* Multi-model nesting rules were over-generalized.

* Model definitions included redundant or superfluous information.

* Simulation frame definition was both incomplete and contained superfluous
information.










Table 3-1 provides a summary of the original version of MXL, since that was the

starting point used in refining the MXL schema.

Table 3-1. List of original MXL elements with descriptions
Element Attributes Children Description
MXL root element

type="xsd:string" defines topology and behavior of a model

type="xsd:string" defines objects and the connectivity of each

object in the model
id="xsd:string"