Building Intelligent Interactive Tutors (eBook)

Front Cover 1
Building Intelligent Interactive Tutors 2
Copyright Page 3
Contents 5
Preface 12
PART I: INTRODUCTION TO ARTIFICIAL INTELLIGENCE AND EDUCATION 14
CHAPTER 1 Introduction 16
1.1 An inflection point in education 17
1.2 Issues addressed by this book 19
1.2.1 Computational issues 20
1.2.2 Professional issues 22
1.3 State of the art in Artificial Intelligence and education 23
1.3.1 Foundations of the field 23
1.3.2 Visions of the field 25
1.3.3 Effective teaching methods 27
1.3.4 Computers in education 29
1.3.5 Intelligent tutors: The formative years 31
1.4 Overview of the book 31
Summary 32
CHAPTER 2 Issues and Features 34
2.1 Examples of intelligent tutors 34
2.1.1 AnimalWatch taught arithmetic 34
2.1.2 PAT taught algebra 37
2.1.3 Cardiac Tutor trained professionals to manage cardiac arrest 40
2.2 Distinguishing features 41
2.3 Learning theories 47
2.3.1 Practical teaching theories 47
2.3.2 Learning theories as the basis for tutor development 49
2.3.3 Constructivist teaching methods 50
2.4 Brief theoretical framework 52
2.5 Computer science, psychology, and education 55
2.6 Building intelligent tutors 57
Summary 58
PART II: REPRESENTATION, REASONING AND ASSESSMENT 60
CHAPTER 3 Student Knowledge 62
3.1 Rationale for building a student model 63
3.2 Basic concepts of student models 63
3.2.1 Domain models 64
3.2.2 Overlay models 65
3.2.3 Bug libraries 65
3.2.4 Bandwidth 66
3.2.5 Open user models 67
3.3 Issues in building student models 68
3.3.1 Representing student knowledge 68
3.3.2 Updating student knowledge 71
3.3.3 Improving tutor performance 72
3.4 Examples of student models 73
3.4.1 Modeling skills: PAT and AnimalWatch 74
3.4.1.1 Pump Algebra Tutor 74
3.4.1.2 AnimalWatch 78
3.4.2 Modeling procedure: The Cardiac Tutor 80
3.4.3 Modeling affect: Affective Learning companions and wayang outpost 82
3.4.3.1 Hardware-based emotion recognition 84
3.4.3.2 Software-based emotion recognition 85
3.4.4 Modeling complex problems: Andes 88
3.5 Techniques to update student models 92
3.5.1 Cognitive science techniques 93
3.5.1.1 Model-tracing tutors 93
3.5.1.2 Constraint-based student model 94
3.5.2 Artificial intelligence techniques 99
3.5.2.1 Formal logic 99
3.5.2.2 Expert-system student models 102
3.5.2.3 Planning and plan-recognition student models 103
3.5.2.4 Bayesian belief networks 105
3.6 Future research issues 106
Summary 107
CHAPTER 4 Teaching Knowledge 108
4.1 Features of teaching knowledge 108
4.2 Teaching models based on human teaching 112
4.2.1 Apprenticeship training 112
4.2.1.1 SOPHIE: An example of apprenticeship training 113
4.2.1.2 Sherlock: An example of an apprenticeship environment 114
4.2.2 Problem solving 116
4.3 Teaching Models informed by learning theory 118
4.3.1 Pragmatics of human learning theories 119
4.3.2 Socratic learning theory 120
4.3.2.1 Basic principles of Socratic learning theory 120
4.3.2.2 Building Socratic tutors 122
4.3.3 Cognitive learning theory 123
4.3.3.1 Basic principles of cognitive learning theories 123
4.3.3.2 Building cognitive learning tutors 123
4.3.3.2.1 Adaptive control of thought (ACT) 124
4.3.3.2.2 Building cognitive tutors 124
4.3.3.2.3 Development and deployment of model-tracing tutors 125
4.3.3.2.4 Advantages and limitations of model-tracing tutors 125
4.3.4 Constructivist theory 127
4.3.4.1 Basic principles of constructivism 127
4.3.4.2 Building constructivist tutors 128
4.3.5 Situated learning 130
4.3.5.1 Basic principles of situated learning 130
4.3.5.2 Building situated tutors 131
4.3.6 Social interaction and zone of proximal development 136
4.3.6.1 Basic principles of social interaction and zone of proximal development 136
4.3.6.2 Building social interaction and ZPD tutors 137
4.4 Teaching models facilitated by technology 139
4.4.1 Features of animated pedagogical agents 140
4.4.2 Building animated pedagogical agents 142
4.4.2.1 Emotive agents 144
4.4.2.2 Life quality 144
4.5 Industrial and Military Training 145
4.6 Encoding multiple teaching strategies 146
Summary 147
CHAPTER 5 Communication Knowledge 149
5.1 Communication and teaching 149
5.2 Graphic communication 151
5.2.1 Synthetic humans 151
5.2.2 Virtual reality environments 155
5.2.3 Sophisticated graphics techniques 162
5.3 Social intelligence 163
5.3.1 Visual recognition of emotion 164
5.3.2 Metabolic indicators 166
5.3.3 Speech cue recognition 168
5.4 Component interfaces 169
5.5 Natural language communication 171
5.5.1 Classification of natural language-based intelligent tutors 171
5.5.1.1 Mixed initiative dialogue 172
5.5.1.2 Single-initiative dialogue 174
5.5.1.3 Directed dialogue 177
5.5.1.4 Finessed dialogue 178
5.5.2 Building natural language tutors 180
5.5.2.1 Basic principles in natural language processing 180
5.5.2.2 Tools for building natural language tutors 182
5.6 Linguistic issues in natural language processing 185
5.6.1 Speech understanding 185
5.6.1.1 LISTEN: The Reading Tutor 186
5.6.1.2 Building speech understanding systems 187
5.6.2 Syntactic processing 188
5.6.3 Semantic and pragmatic processing 190
5.6.4 Discourse processing 192
Summary 194
CHAPTER 6 Evaluation 196
6.1 Principles of intelligent tutor evaluation 196
6.1.1 Establish goals of the tutor 197
6.1.2 Identify goals of the evaluation 197
6.1.3 Develop an evaluation design 201
6.1.3.1 Build an evaluation methodology 201
6.1.3.2 Consider alternative evaluation comparisons 204
6.1.3.3 Outline the evaluation design 206
6.1.4 Instantiate the evaluation design 209
6.1.4.1 Consider the variables 209
6.1.4.2 Select target populations 210
6.1.4.3 Select control measures 210
6.1.4.4 Measure usability 211
6.1.5 Present results 211
6.1.6 Discuss the evaluation 213
6.2 Example of intelligent tutor evaluations 213
6.2.1 Sherlock: A tutor for complex procedural skills 213
6.2.2 Stat Lady: A statistics tutor 215
6.2.3 LISP and PAT: Model tracing tutors 217
6.2.4 Database tutors 222
6.2.5 Andes: A physics tutor 225
6.2.6 Reading Tutor: A tutor that listens 228
6.2.7 AnimalWatch: An arithmetic tutor 230
Summary 233
PART III: TECHNOLOGIES AND ENVIRONMENTS 234
CHAPTER 7 Machine Learning 236
7.1 Motivation for machine learning 236
7.2 Building machine learning techniques into intelligent tutors 241
7.2.1 Machine learning components 241
7.2.2 Supervised and unsupervised learning 243
7.3 Features learned by intelligent tutors using machine learning techniques 245
7.3.1 Expand student and domain models 245
7.3.2 Identify student learning strategies 247
7.3.3 Detect student affect 248
7.3.4 Predict student performance 248
7.3.5 Make teaching decisions 249
7.4 Machine learning techniques 252
7.4.1 Uncertainty in tutoring systems 252
7.4.1.1 Basic probability notation 254
7.4.1.2 Belief networks in tutors 255
7.4.2 Bayesian belief networks 257
7.4.2.1 Bayesian belief networks in intelligent tutors 260
7.4.2.2 Examples of Bayesian student models 261
7.4.2.2.1 Expert-centric Bayesian models 262
7.4.2.2.2 Data-centric Bayesian models 266
7.4.2.2.3 Efficiency-centric Bayesian models 267
7.4.2.3 Building Bayesian belief networks 268
7.4.2.3.1 Define the structure of the Bayesian network 268
7.4.2.3.2 Initialize values in a Bayesian network 270
7.4.2.3.3 Update probabilities in a Bayesian network 271
7.4.2.4 Advantages of Bayesian networks and comparison with model-based tutors 276
7.4.3 Reinforcement learning 277
7.4.3.1 Examples of reinforcement learning 278
7.4.3.2 Building reinforcement learners 279
7.4.3.3 Reinforcement learning in intelligent tutors 280
7.4.3.4 Animal learning and reinforcement learning 281
7.4.4 Hidden Markov models 282
7.4.5 Decision theoretic reasoning 287
7.4.6 Fuzzy logic 292
7.5 Examples of intelligent tutors that employ machine learning techniques 294
7.5.1 Andes: Bayesian belief networks to reason about student knowledge 294
7.5.1.1 Sources of uncertainty and structure of the Andes-Bayesian network 294
7.5.1.2 Infer student knowledge 296
7.5.1.3 Self-Explain Tutor 299
7.5.1.4 Limitations of the Andes Bayesian networks 302
7.5.2 AnimalWatch: Reinforcement learning to predict student actions 302
7.5.2.1 Reinforcement learning in AnimalWatch 303
7.5.2.2 Gather training data for the machine learner 305
7.5.2.3 Induction techniques used by the learning mechanism 306
7.5.2.4 Evaluation of the reinforcement learning tutor 306
7.5.2.5 Limitations of the AnimalWatch reinforcement learner 309
Summary 310
CHAPTER 8 Collaborative Inquiry Tutors 311
8.1 Motivation and research issues 311
8.2 Inquiry Learning 312
8.2.1 Benefits and challenges of inquiry-based learning 313
8.2.2 Three levels of inquiry support 315
8.2.2.1 Tools that structure inquiry 315
8.2.2.2 Tools that monitor inquiry 318
8.2.2.3 Tools that offer advice 320
8.2.2.3.1 Belvedere 321
8.2.2.3.2 Rashi 323
8.2.3 Phases of the inquiry cycle 328
8.3 Collaborative Learning 329
8.3.1 Benefits and challenges of collaboration 330
8.3.2 Four levels of collaboration support 332
8.3.2.1 Tools that structure collaboration 333
8.3.2.2 Tools that mirror collaboration 334
8.3.2.3 Tools that provide metacognitive support 337
8.3.2.4 Tools that coach students in collaboration 343
8.3.3 Phases of Collaboration 346
8.4 Summary and discussion 348
CHAPTER 9 Web-Based Learning Environments 350
9.1 Educational inflection point 350
9.2 Conceptual framework for Web-based learning 353
9.3 Limitation of Web-based instruction 356
9.4 Variety of Web-based resources 357
9.4.1 Adaptive systems 358
9.4.1.1 Example of an adaptive system 359
9.4.1.2 Building iMANIC 360
9.4.1.3 Building adaptive systems 364
9.4.1.3.1 Adaptive navigation: Customize travel to new pages 364
9.4.1.3.2 Adaptive Presentation: Customize page content 367
9.4.2 Tutors ported to the Web 368
9.5 Building the Internet 369
9.6 Standards for Web-based resources 372
9.7 Education Space 374
9.7.1 Education Space: Services description 376
9.7.2 Education Space: Nuts and bolts 378
9.7.2.1 Semantic Web 379
9.7.2.2 Ontologies 382
9.7.2.3 Agents and networking issues 385
9.7.2.4 Teaching Grid 386
9.8 Challenges and technical issues 387
9.9 Vision of the Internet 390
Summary 391
CHAPTER 10 Future View 393
10.1 Perspectives on educational futures 393
10.1.1 Political and social viewpoint 394
10.1.2 Psychological perspective 396
10.1.3 Classroom teachers' perspective 397
10.2 Computational vision for education 399
10.2.1 Hardware and software development 399
10.2.2 Artificial intelligence 401
10.2.3 Networking, mobile, and ubiquitous computing 402
10.2.4 Databases 405
10.2.5 Human-computer interfaces 406
10.3 Where are all the intelligent tutors? 407
10.3.1 Example authoring tools 408
10.3.2 Design tradeoffs 411
10.3.3 Requirements for building intelligent tutor authoring tools 412
10.4 Where are we going? 414
References 416
Index 464
A 464
B 465
C 466
D 467
E 468
F 469
G 469
H 469
I 470
J 471
K 471
L 472
M 473
N 474
O 474
P 475
Q 476
R 476
S 476
T 478
U 480
V 480
W 480
X 480
Z 480