User Modelling in Text Generation

✍ Scribed by Cécile L. Paris

• Publisher: Bloomsbury Academic
• Year: 2015
• Tongue: English
• Leaves: 242
• Series: Linguistics: Bloomsbury Academic Collections
• Category: Library

✦ Synopsis

This book addresses the issue of how the user's level of domain knowledge affects interaction with a computer system. It demonstrates the feasibility of incorporating a model of user's domain knowledge into a natural language generation system.

✦ Table of Contents

Cover
Half-title
Title
Copyright
Dedication
Contents
List of Figures
Foreword
Preface
1 Introduction
1.1 Natural Language Generation
1.1.1 Language variation according to the situation
1.2 User modelling in generation
1.3 Research method and major results
1.4 The domain
1.5 System overview
1.6 Examples from TAILOR
1.7 Limitations
1.8 A guide to remaining chapters
2 The user model in TAILOR
2.1 Contents and use of a user model
2.2 Identifying what needs to be in the user model
2.3 Determining the level of expertise
2.3.1 User type
2.3.2 Obtaining information from a dialogue
2.3.3 Employing Inference Rules
2.3.4 Asking the user questions and using the previous discourse
2.4 Conclusions
3 The text analysis
3.1 Analyzing text
3.2 The texts analyzed
3.2.1 The textual analysis
3.2.2 Analyses of entries from adult encyclopedias and the car manual for experts
3.2.3 Texts from junior encyclopedias, high school textbooks, and the car manual for novices
3.2.4 Need for directives
3.2.5 Summary of the textual analyses
3.2.6 Plausibility of this hypothesis
3.3 Combining the strategies for users with intermediate levels of expertise
3.4 Summary
4 The discourse strategies used in TAILOR
4.1 Constituency Schema
4.2 Process Trace
4.3 Requirements of the knowledge base
4.4 The process trace: a procedural strategy
4.4.1 Identifying the main path and different kinds of links
4.4.2 The main path
4.4.3 Deciding among several side links or side chains
4.5 Strategy representation
4.6 Open problems
4.7 Summary
5 Combining the strategies to describe devices for a wholerange of users
5.1 The user model contains explicit parameters
5.2 Generating a description based on the user model
5.2.1 Choosing a strategy for the overall structure of the description
5.2.2 Combining the strategies
5.3 Examples of texts combining the two strategies
5.4 Combining strategies yields a greater variety of texts
5.5 Conclusions
6 TAILOR system implementation
6.1 System overview
6.2 The knowledge base and its representation
6.2.1 The generalization hierarchies
6.2.2 Limitations of the knowledge base
6.3 The user model
6.4 The textual component
6.4.1 Initially selecting a strategy
6.4.2 Finding the main path
6.4.3 Implementation of the Strategies
6.4.4 Stepping through the Constituency Schema
6,4.5 The ATN corresponding to the Process Trace
6.4.6 Choosing an arc
6.5 The Interface
6.6 The surface generator
6.6.1 The functional grammar and the unification process
6.7 Issues pertaining to domain dependency
6.8 TAILOR as a question answering system
7 Related Work
7.1 Related work in user modelling and generation
7.1.1 Using stereotypes
7.1.2 Modelling and exploiting the user's domain knowledge
7.1.3 Using knowledge about the user's plans and goals to generate responses
7.1.4 Using reasoning about mutual beliefs to plan an utterance
7.1.5 Dealing with misconceptions about the domain
7.1.6 Exploiting the speaker's and hearer's 'pragmatic goals' in generation
7.2 Related work in psychology and reading comprehension
7.3 Summary
8 Beyond TAILOR
8.1 Main points of this work
8.2 Feasibility and extensibility of this approach
8.3 Current directions
8.3.1 Employing the generator presented here in another domain
8.3.2 Building a system with dialogue capabilities
8.3.3 Tailoring the phrasing of a text
83.4 Bringing the whole generation process under register control
8.3.5 Providing a system with adaptive capabilities
Appendix A
Appendix B
Bibliography
Index