B.C. Papegaaij

V. Sadler and A.P.M. Witkam (eds.)

WORD EXPERT SEMANTICS
an Interlingual
Knowledge-Based
Approach

 

1986

FORIS PUBLICATIONS

Dordrecht - Holland/Riverton - U.S.A.

 

(Distributed Language Translation 1)

 

[copyright: Mouton de Gruyter]

 

Contents

 

Introduction 1

Part I - Machine Translation and the Problem of Ambiguity 4

Chapter 1 Natural Language Processing 5

-      The Computer (R)evolution 5

-      Natural Language 5

-      Machine Translation 6

-      Limitations of the Computer 9

-      Natural Language versus the Computer 10

Chapter 2 Lexical Ambiguity 19

-      Paradigms of Meaning 19

-      Non-determinism 22

-      Knowledge of the World 25

-      The Shiftiness and Creativity of Language 26

Chapter 3 A Concise Historical Survey 30

-      Word-by-Word Translation 30

-      Syntax Only 32

-      Closed Worlds 32

-      Selection Restrictions 36

-      Case Grammar 38

-      Scripts and Plans 39

-      Preference Semantics 43

-      The Word Expert 46

Chapter 4 MT & Understanding - Contemporary Techniques 48

-      'Decompose and Capture' - Classification by Means of Primitives 48

-      The Knowledge Bank - Storing Facts and Figures 51

-      Sublanguage & Limited Domains 53

-      Semantic Networks 54

-      Frames 57

-      Logic, Inferencing & Mechanical Reasoning 60

Chapter 5 The Importance of the Dictionary 65

-      The Lexical Knowledge Bank 66

-      Lexical Universe & Lexical Function 68

-      Innovations in Lexicography 71

Chapter 6 The Need for Integration 73

-      Many Solutions for Many Problems 73

-      High Quality MT through Combined Strategies 73

Part II - The Semantic Word Expert System 75

Chapter 1 General Lay-out of the DLT System 77

-      Modular Approach 77

-      On-line Parsing 79

-      Interleaved Semantic Processing 81

-      End-of-Sentence Macrocontext Semantics 82

-      Disambiguation Dialogue 84

Chapter 2 The Structure of the Lexical Knowledge Bank 85

-      Principal Structure 85

-      SL to lL 86

-      IL to TL 88

-      The Central Role of IL 89

-      The Lexical Taxonomy 91

-      Contextual Dependency: SOLL Pairs 95

-      Logical & Ontological Information 96

-      The Core 99

-      Relator Matching 100

-      Closed Sets, Cyclic Data and other Information 102

Chapter 3 Disambiguation with SWESIL 105

-      Syntactic Analysis 105

-      Interleaved Semantics 108

-      The IST Pair Generator 108

-      Paradigmatic Expansion 112

-      The Matching Cycle 113

-      Ordering the Alternatives 127
Second-Order Matching 127

Chapter 4 The Disambiguation Dialogue 129

-       Closed Format Question 131

-       Types of Questions & On-screen Presentation 133

-       Experiments with Disambiguation Dialogues 136

Part III - The Semantic Work Bench - A Development Tool 143

Chapter 1 Developing SWESIL and the LKB 144

-       Building a Lexicon 144

-       Experimental Lexicographic Approaches 147

-       The Melby Test 152

-       The Size of the LKB 156

Chapter 2 The Semantic Work Bench 163

-       The Experimental Environment 163

-       Modular Approach 164

-       Tools 166

Chapter 3 Tests and Results 170

-       Single IST Pairs 170

-       Sentences with First-Order Matching Only 174

-       Sentences with Second-Order Matching 188

Chapter 4 Melby Test Results 191

Part IV - Future Developments 196

Chapter 1 Computerized Lexicography 197

-       Diversity of Sources 198

-       Formalizing Information 200

-       Storing Entries in the LKB 201

-       Lexicographic Tools 202

-       The Ultimate Aim - The General Purpose IL Knowledge Bank 205

Chapter 2 Macrocontext and Discourse Analysis 208

-       Beyond the Sentence Boundary 208

-       Text Models 210

-       Incremental Understanding - Validation of Information from Different Sources 211

-       Applications - Summaries, Indexing, Rephrasing 213

-       Increasing Dialogue Efficiency 213


Chapter 3 The Self-Improving System 215

-       Changing Language 215

-       Variety of Language 216

-       Learning from Experience Improving Performance while Processing Text 217

-       The Dialogue as SWESIL's Tutor 220

-       Updating and Expanding 221

References 222