Spoken Natural Language Dialog Systems : A Practical Approach.

Intro -- Contents -- 1 Achieving Spoken Communication with Computers -- 1.1 Problem Solving Environment: Task-Oriented Dialogs -- 1.2 Integrating Dialog with Task Assistance: The Target Behaviors -- 1.2.1 Problem Solving to Achieve a Goal -- 1.2.2 Subdialogs and Effective Movement Between Them -- 1.2.3 Accounting for User Knowledge and Abilities -- 1.2.4 Expectation of User Input -- 1.2.5 Variable Initiative -- 1.2.6 Integrated Behavior Via the Missing Axiom Theory -- 1.3 Preliminary Study -- 1.4 An Outline of the Book -- 2 Foundational Work in Integrated Dialog Processing -- 2.1 Problem Solving in an Interactive Environment -- 2.2 Language Use in a Problem-Solving Environment -- 2.2.1 The Missing Axiom Theory -- 2.2.2 Speech Act Theory -- 2.2.3 Computational Speech Act Theory: Analyzing Intentions -- 2.2.4 Differing Subdialog Purposes: The Plan-Based Theory of Litman and Allen -- 2.2.5 Collective Intentions -- 2.3 User Model -- 2.3.1 General User Modeling Architecture -- 2.3.2 Using User Model Information in Generation -- 2.3.3 Acquiring User Model Information -- 2.4 Expectation Usage -- 2.4.1 Speech Recognition -- 2.4.2 Plan Recognition -- 2.5 Variable Initiative Theory -- 2.5.1 Defining Initiative -- 2.5.2 Discourse Structure in Variable Initiative Dialogs -- 2.5.3 Plan Recognition for Variable Initiative Dialog -- 2.6 Integrated Dialog Processing Theory -- 2.6.1 Subdialog Switching: Reichman's Conversational Moves -- 2.6.2 Beyond Speech Acts: Conversation Acts of Traum and Hinkelman -- 2.6.3 Integrated Discourse Structure: The Tripartite Model of Grosz and Sidner -- 2.7 Dialog Systems -- 2.7.1 Requirements -- 2.7.2 Portable Systems -- 2.7.3 Question-Answer Systems: Keyboard Input -- 2.7.4 Spoken Input Systems -- 2.7.5 A Discourse System -- 2.7.6 Variable Initiative Systems -- 2.8 Summary -- 3 Dialog Processing Theory.

3.1 System Architecture -- 3.2 Modeling Interactive Task Processing -- 3.2.1 Computer and User Prerequisites -- 3.2.2 A Domain-Independent Language for Describing Goals, Actions, and States -- 3.2.3 Robust Selection of Task Steps -- 3.2.4 Determining Task Step Completion -- 3.2.5 What About Dialog? -- 3.3 Integrating Task Processing with Dialog: The Missing Axiom Theory -- 3.3.1 The Role of Language: Supplying Missing Axioms -- 3.3.2 Interruptible Theorem Proving Required &amp -- #8658 -- IPSIM -- 3.4 Exploiting Dialog Context: User Model -- 3.4.1 Accounting for User Knowledge and Abilities -- 3.4.2 Computing Inferences from User Input -- 3.4.3 User Model Usage: Integrating Task Processing with Dialog -- 3.5 Exploiting Dialog Context: Input Expectations -- 3.5.1 Foundations of Expectation-Driven Processing -- 3.5.2 Using Expectation-Driven Processing -- 3.6 A Theory of Variable Initiative Dialog -- 3.6.1 Defining Variable Initiative and Dialog Mode -- 3.6.2 Response Formulation in Variable Initiative Dialog -- 3.7 Putting the Pieces Together -- 3.7.1 What Is a Dialog? -- 3.7.2 Integrated Theory -- 4 Computational Model -- 4.1 Dialog Processing Algorithm -- 4.1.1 Motivation and Basic Steps -- 4.1.2 Tracing the Basic Steps -- 4.2 Receiving Suggestion from Domain Processor -- 4.3 Selection of Next Goal -- 4.4 Attempting Goal Completion -- 4.4.1 Step 2a: Attempt to Prove Completion -- 4.4.2 Step 2b: Computing Final Utterance Specification -- 4.4.3 Step 2c: Computing Expectations for the User's Response -- 4.4.4 Step 2d: Receiving User Input -- 4.4.5 Step 2e: Computing World Interpretation -- 4.4.6 Steps 2f and 2g: Updating Context and Discourse Structure -- 4.4.7 Step 2h: Computing Inferences from the Input -- 4.4.8 Step 2i: Selecting Applicable Axiom -- 4.5 Updating System Knowledge -- 4.6 Determine Next Domain Processor Operation.

4.7 Solutions to Dialog Processing Problems -- 4.7.1 Interrupts -- 4.7.2 Robustness and the Handling of Speech Recognition Errors -- 4.7.3 Variable Initiative Dialog -- 4.8 Integrated Dialog Processing: A Summary -- 5 Parsing -- 5.1 Introduction -- 5.2 Overview of the Parser -- 5.3 The Parser Input Lattice -- 5.3.1 What is in a Word? -- 5.3.2 Uncertain Inputs -- 5.3.3 Arc Weights -- 5.3.4 Indexing Lattice Nodes -- 5.3.5 Inputs Used in the Experiments -- 5.4 Translation Grammars -- 5.5 Minimum Distance Translation -- 5.5.1 Distance Between Strings -- 5.5.2 A Precise Definition of What the MDT Algorithm Does -- 5.6 An Efficient Algorithm for MDT -- 5.6.1 Data Structures Used by MDT -- 5.6.2 The Outer Procedure -- 5.6.3 The Inner Procedure -- 5.6.4 An Important Optimization -- 5.7 Enhancements to the MDT Algorithm -- 5.7.1 Lexicon Dependent Deletion and Insertion Costs -- 5.7.2 Grammar Dependent Insertion Costs -- 5.8 Expectation Processing -- 5.8.1 Wildcards -- 5.8.2 Wildcard String Matching -- 5.8.3 Enhancements to the Minimum Matching String Algorithm -- 5.8.4 Wildcard String Matching Versus Unification -- 5.8.5 Expectation Based Hypothesis Selection -- 5.8.6 The Expectation Function -- 5.9 Computational Complexity -- 5.9.1 Notation -- 5.9.2 The Complexity of Input Lattice Node Renumbering -- 5.9.3 The Complexity of MDT -- 5.9.4 The Complexity of Expectation Processing -- 5.9.5 Overall Parser Complexity -- 6 System Implementation -- 6.1 Knowledge Representation -- 6.1.1 Prolog -- 6.1.2 GADL -- 6.1.3 snf -- 6.1.4 Sef -- 6.1.5 IPSIM -- 6.1.6 Discourse Structure -- 6.1.7 Axioms -- 6.1.8 Interfaces -- 6.2 Domain Processor -- 6.2.1 Debugging Methodology -- 6.2.2 Decision Making Strategies -- 6.2.3 Debugging Control Strategy Modifications for Dialog -- 6.3 Generation -- 6.3.1 Overview -- 6.3.2 Natural Language Directions for Locating Objects.

6.4 Resource Utilization -- 7 Experimental Results -- 7.1 Hypotheses -- 7.2 Preliminary Results -- 7.3 Experimental Design -- 7.3.1 Overview -- 7.3.2 Problem Selection -- 7.3.3 Session 1 Procedure -- 7.3.4 Session 2 Procedure -- 7.3.5 Session 3 Procedure -- 7.4 Experimental Setup -- 7.5 Subject Pool -- 7.6 Cumulative Results -- 7.6.1 Basic System Performance -- 7.6.2 Parameter Definitions -- 7.6.3 Aggregate Results -- 7.6.4 Results as a Function of Problem -- 7.6.5 Statistical Analysis of the Results -- 7.7 Results from Subject Responses about System Usage -- 7.8 Conclusions -- 8 Performance of the Speech Recognizer and Parser -- 8.1 Preparation of the Data -- 8.2 Speech Recognizer Performance -- 8.2.1 Comparison to Other Speech Recognizers -- 8.2.2 Comparison to Humans -- 8.3 Parser Performance -- 8.4 Optimal Expectation Functions -- 9 Enhanced Dialog Processing: Verifying Doubtful Inputs -- 9.1 Handling Misunderstandings -- 9.2 Deciding When to Verify -- 9.2.1 Confidence Estimates -- 9.2.2 Selecting a Verification Threshold -- 9.3 Experimental Results -- 9.4 Summary of Verification Subdialogs -- 10 Extending the State of the Art -- 10.1 Continuing Work -- 10.1.1 Automatic Switching of Initiative -- 10.1.2 Exploiting Dialog Context in Response Generation -- 10.1.3 Miscommunication and Metadialog -- 10.1.4 Less Restricted Vocabulary -- 10.1.5 Evaluating Model Applicability -- 10.2 Where Do We Go Next? -- A: The Goal and Action Description Language -- B: User's Guide for the Interruptible Prolog SIMulator (IPSIM) -- B.1 Introduction -- B.2 Specifying Rules and Axioms for IPSIM -- B.2.1 Sample Specification and Description -- B.2.2 Additional Requirements for the Specification -- B.2.3 The Special Clauses of IPSIM -- B.3 Using IPSIM -- B.3.1 The IPSIM Command Language -- B.3.2 The Use of Knowledge -- B.3.3 A Sample Control Scheme.

B.4 Creating Dynamic Lists of Missing Axioms -- B.4.1 The Defaults -- B.4.2 Redefining axiom-need -- B.5 Using IPSIM Calls within Theorem Specifications -- C: Obtaining the System Software Via Anonymous FTP -- Bibliography -- Index -- A -- B -- C -- D -- E -- F -- G -- H -- I -- J -- K -- L -- M -- N -- O -- P -- Q -- R -- S -- T -- U -- V -- W -- Y -- Z.

Description based on publisher supplied metadata and other sources.

Electronic reproduction. Ann Arbor, Michigan : ProQuest Ebook Central, 2024. Available via World Wide Web. Access may be limited to ProQuest Ebook Central affiliated libraries.