Spoken Natural Language Dialog Systems A Practical Approach

by Smith, Ronnie W.; Hipp, D. Richard

• ISBN13: 9780195091878
• ISBN10: 0195091876
• Format: Hardcover
• Copyright: 1995-02-02
• Publisher: Oxford University Press

Summary

As spoken natural language dialog systems technology continues to make great strides, numerous issues regarding dialog processing still need to be resolved. This book presents an exciting new dialog processing architecture that allows for a number of behaviors required for effective human-machine interactions, including: problem-solving to help the user carry out a task, coherent subdialog movement during the problem-solving process, user model usage, expectation usage for contextual interpretation and error correction, and variable initiative behavior for interacting with users of differing expertise. The book also details how different dialog problems in processing can be handled simultaneously, and provides instructions and in-depth result from pertinent experiments. Researchers and professionals in natural language systems will find this important new book an invaluable addition to their libraries.

Table of Contents

Achieving Spoken Communication with Computers	p. 3
Problem Solving Environment: Task-Oriented Dialogs	p. 6
Integrating Dialog with Task Assistance: The Target Behaviors	p. 7
Problem Solving to Achieve a Goal	p. 8
Subdialogs and Effective Movement Between Them	p. 8
Accounting for User Knowledge and Abilities	p. 10
Expectation of User Input	p. 11
Variable Initiative	p. 11
Integrated Behavior Via the Missing Axiom Theory	p. 12
Preliminary Study	p. 13
An Outline of the Book	p. 13
Foundational Work in Integrated Dialog Processing	p. 15
Problem Solving in an Interactive Environment	p. 15
Language Use in a Problem-Solving Environment	p. 16
The Missing Axiom Theory	p. 16
Speech Act Theory	p. 17
Computational Speech Act Theory: Analyzing Intentions	p. 18
Differing Subdialog Purposes: The Plan-Based Theory of Litman and Allen	p. 21
Collective Intentions	p. 22
User Model	p. 23
General User Modeling Architecture	p. 24
Using User Model Information in Generation	p. 26
Acquiring User Model Information	p. 27
Expectation Usage	p. 29
Speech Recognition	p. 29
Plan Recognition	p. 29
Variable Initiative Theory	p. 31
Defining Initiative	p. 31
Discourse Structure in Variable Initiative Dialogs	p. 32
Plan Recognition for Variable Initiative Dialog	p. 32
Integrated Dialog Processing Theory	p. 33
Subdialog Switching: Reichman's Conversational Moves	p. 33
Beyond Speech Acts: Conversation Acts of Traum and Hinkelman	p. 35
Integrated Discourse Structure: The Tripartite Model of Grosz and Sidner	p. 36
Dialog Systems	p. 38
Requirements	p. 39
Portable Systems	p. 39
Question-Answer Systems: Keyboard Input	p. 42
Spoken Input Systems	p. 42
A Discourse System	p. 44
Variable Initiative Systems	p. 45
Summary	p. 46
Dialog Processing Theory	p. 47
System Architecture	p. 47
Modeling Interactive Task Processing	p. 51
Computer and User Prerequisites	p. 51
A Domain-Independent Language for Describing Goals, Actions, and States	p. 52
Robust Selection of Task Steps	p. 54
Determining Task Step Completion	p. 55
What About Dialog?	p. 57
Integrating Task Processing with Dialog: The Missing Axiom Theory	p. 57
The Role of Language: Supplying Missing Axioms	p. 58
Interruptible Theorem Proving Required [implies] IPSIM	p. 58
Exploiting Dialog Context: User Model	p. 59
Accounting for User Knowledge and Abilities	p. 59
Computing Inferences from User Input	p. 60
User Model Usage: Integrating Task Processing with Dialog	p. 60
Exploiting Dialog Context: Input Expectations	p. 63
Foundations of Expectation-Driven Processing	p. 63
Using Expectation-Driven Processing	p. 64
A Theory of Variable Initiative Dialog	p. 68
Defining Variable Initiative and Dialog Mode	p. 68
Response Formulation in Variable Initiative Dialog	p. 70
Putting the Pieces Together	p. 72
What Is a Dialog?	p. 72
Integrated Theory	p. 73
Computational Model	p. 75
Dialog Processing Algorithm	p. 75
Motivation and Basic Steps	p. 75
Tracing the Basic Steps	p. 77
Receiving Suggestion from Domain Processor	p. 78
Selection of Next Goal	p. 79
Attempting Goal Completion	p. 81
Step 2a: Attempt to Prove Completion	p. 87
Step 2b: Computing Final Utterance Specification	p. 88
Step 2c: Computing Expectations for the User's Response	p. 89
Step 2d: Receiving User Input	p. 94
Step 2e: Computing World Interpretation	p. 95
Steps 2f and 2g: Updating Context and Discourse Structure	p. 96
Step 2h: Computing Inferences from the Input	p. 97
Step 2i: Selecting Applicable Axiom	p. 97
Updating System Knowledge	p. 101
Determine Next Domain Processor Operation	p. 102
Solutions to Dialog Processing Problems	p. 103
Interrupts	p. 103
Robustness and the Handling of Speech Recognition Errors	p. 115
Variable Initiative Dialog	p. 117
Integrated Dialog Processing: A Summary	p. 119
Parsing	p. 121
Introduction	p. 121
Overview of the Parser	p. 123
The Parser Input Lattice	p. 125
What is in a Word?	p. 125
Uncertain Inputs	p. 126
Arc Weights	p. 127
Indexing Lattice Nodes	p. 128
Inputs Used in the Experiments	p. 129
Translation Grammars	p. 130
Minimum Distance Translation	p. 132
Distance Between Strings	p. 132
A Precise Definition of What the MDT Algorithm Does	p. 133
An Efficient Algorithm for MDT	p. 135
Data Structures Used by MDT	p. 135
The Outer Procedure	p. 136
The Inner Procedure	p. 137
An Important Optimization	p. 141
Enhancements to the MDT Algorithm	p. 142
Lexicon Dependent Deletion and Insertion Costs	p. 142
Grammar Dependent Insertion Costs	p. 143
Expectation Processing	p. 144
Wildcards	p. 144
Wildcard String Matching	p. 145
Enhancements to the Minimum Matching String Algorithm	p. 148
Wildcard String Matching Versus Unification	p. 149
Expectation Based Hypothesis Selection	p. 149
The Expectation Function	p. 149
Computational Complexity	p. 151
Notation	p. 151
The Complexity of Input Lattice Node Renumbering	p. 151
The Complexity of MDT	p. 151
The Complexity of Expectation Processing	p. 153
Overall Parser Complexity	p. 153
System Implementation	p. 155
Knowledge Representation	p. 156
Prolog	p. 156
GADL	p. 156
snf	p. 156
Sef	p. 156
IPSIM	p. 157
Discourse Structure	p. 158
Axioms	p. 159
Interfaces	p. 160
Domain Processor	p. 160
Debugging Methodology	p. 161
Decision Making Strategies	p. 165
Debugging Control Strategy Modifications for Dialog	p. 170
Generation	p. 178
Overview	p. 178
Natural Language Directions for Locating Objects	p. 178
Resource Utilization	p. 179
Experimental Results	p. 181
Hypotheses	p. 181
Preliminary Results	p. 181
Experimental Design	p. 184
Overview	p. 184
Problem Selection	p. 185
Session 1 Procedure	p. 186
Session 2 Procedure	p. 190
Session 3 Procedure	p. 192
Experimental Setup	p. 192
Subject Pool	p. 196
Cumulative Results	p. 197
Basic System Performance	p. 197
Parameter Definitions	p. 197
Aggregate Results	p. 199
Results as a Function of Problem	p. 206
Statistical Analysis of the Results	p. 210
Results from Subject Responses about System Usage	p. 212
Conclusions	p. 214
Performance of the Speech Recognizer and Parser	p. 219
Preparation of the Data	p. 219
Speech Recognizer Performance	p. 221
Comparison to Other Speech Recognizers	p. 223
Comparison to Humans	p. 223
Parser Performance	p. 224
Optimal Expectation Functions	p. 227
Enhanced Dialog Processing: Verifying Doubtful Inputs	p. 231
Handling Misunderstandings	p. 231
Deciding When to Verify	p. 232
Confidence Estimates	p. 232
Selecting a Verification Threshold	p. 237
Experimental Results	p. 238
Summary of Verification Subdialogs	p. 239
Extending the State of the Art	p. 241
Continuing Work	p. 241
Automatic Switching of Initiative	p. 241
Exploiting Dialog Context in Response Generation	p. 242
Miscommunication and Metadialog	p. 244
Less Restricted Vocabulary	p. 245
Evaluating Model Applicability	p. 246
Where Do We Go Next?	p. 247
The Goal and Action Description Language	p. 249
User's Guide for the Interruptible Prolog SIMulator (IPSIM)	p. 253
Introduction	p. 253
Specifying Rules and Axioms for IPSIM	p. 253
Sample Specification and Description	p. 254
Additional Requirements for the Specification	p. 254
The Special Clauses of IPSIM	p. 256
Using IPSIM	p. 257
The IPSIM Command Language	p. 257
The Use of Knowledge	p. 262
A Sample Control Scheme	p. 262
Creating Dynamic Lists of Missing Axioms	p. 262
The Defaults	p. 262
Redefining axiom_need	p. 262
Using IPSIM Calls within Theorem Specifications	p. 264
Obtaining the System Software Via Anonymous FTP	p. 265
Bibliography	p. 267
Index	p. 279
Table of Contents provided by Syndetics. All Rights Reserved.