دانلود پایان نامه دکترا به زبان انگلیسی Lexical semantics and knowledge representation in multilingual sentence generation

اختصاصی از یارا فایل دانلود پایان نامه دکترا به زبان انگلیسی Lexical semantics and knowledge representation in multilingual sentence generation دانلود با لینک مستقیم و پرسرعت .

دانلود پایان نامه دکترا به زبان انگلیسی Lexical semantics and knowledge representation in multilingual sentence generation که شامل 170 صفحه و بشرح زیر میباشد:

نوع فایل : PDF قابل ویرایش

Abstract
Lexical semantics and knowledge representation in multilingual sentence generation
Manfred Stede
Doctor of Philosophy Graduate Department of Computer Science University of Toronto 1996
This thesis develops a new approach to automatic language generation that focuses on the need
to produce a range of dierent paraphrases from the same input representation. One novelty of
the system is its solidly grounding representations of word meaning in a background knowledge
base, which enables the production of paraphrases stemming from certain inferences, rather
than from purely lexical relationships alone.
The system is designed in such a way that the paraphrasing mechanism extends naturally to
a multilingual generator; specifically, we will be concerned with producing English and German
sentences. The focus of the system is on lexical paraphrases, and one of the contributions of the
thesis is in identifying, analyzing and extending relevant linguistic research so that it can be
used to handle the problems of lexical semantics in a language generation system. The lexical
entries are more complex than in previous generators, and they separate the various aspects
of word meaning, so that dierent ways of paraphrasing can be systematically related to the
dierent motivations for saying a sentence in a particular way. One result of accounting for
lexical semantics in this fashion is a formalization of a number of verb alternations, for which
a generative treatment is given.
While the actual choice of one paraphrase as the best-suited utterance in a given situation is
not a focal point of the thesis, two dimensions of preferring a variant of a sentence are discussed:
that of assigning salience to the dierent elements of the sentence, and that of connotational or
stylistic features of the utterance. These dimensions are integrated into the system, and it can
thus determine a preferred paraphrase from a set of alternatives.
To demonstrate the feasibility of the approach, the proposed generation architecture has
been implemented as a protoype, along with a domain model that serves as the background
knowledge base for specifying the input to the generator. A range of generated examples is
...presented to show the functionality of the system

Contents
1 Introduction 1
1.1 Natural language generation . 1
1.2 Background: the TECHDOC generator . 3
1.3 Goals of this research .. 4
1.4 Overview of the research and its results . 5
1.5 Organization of the thesis . : 10
2 Lexicalization in NLG 12
2.1 Introduction .. 12
2.2 The nature of lexical items in NLP. 13
2.3 Criteria for lexical choice . : 14
2.3.1 Salience .. : 15
2.3.2 Pragmatics and style . 16
2.4 Linking concepts to lexical items. : 17
2.4.1 Discrimination nets . 17
2.4.2 Taxonomic knowledge bases and the lexicon.. 18
2.5 Placing lexicalization in the generation process . : 20
2.5.1 Lexical and other choices. : 20
2.5.2 PENMAN .. 21
2.6 Multilingual generation .. 23
2.7 Conclusions: making progress on lexicalization . : 23
3 Lexical semantics 27
3.1 Introduction .. 27
3.2 Relational theories of word meaning .. 28
3.3 Decomposition .. : 29
3.4 Denotation versus connotation.. 31
3.5 Two-level semantics .. 32
3.6 Aspect and Aktionsart .. 34
3.7 Valency and case frames .. 36
3.8 Verb alternations .. 37
3.9 Salience . . 39
3.10 Conclusions: word meaning in NLG .. 41
4 Classifying lexical variation 44
4.1 Intra-lingual paraphrases .. 44
4.2 Inter-lingual divergences .. 47
4.3 Divergences as paraphrases . 49
5 Modelling the domain 51
5.1 Building domain models for NLG. : 51
5.2 Background: knowledge representation in LOOM . 53
5.3 Ontological categories in our system .. 54
5.4 A domain model for containers and liquids . : 58
5.4.1 Objects .. : 60
5.4.2 Qualities .. 60
5.4.3 States ... 60
5.4.4 Activities .. 65
5.4.5 Events ... 65
6 Levels of representation: SitSpec and SemSpec 67
6.1 Finding appropriate levels of representation in NLG. : 67
6.1.1 Decision-making in sentence generation . : 68
6.1.2 A two-level approach . 70
6.2 Linguistic ontology: adapting the 'Upper Model' . 72
6.3 SitSpecs . . 75
6.4 SemSpecs .. : 78
7 Representing the meaning of words: a new synthesis 81
7.1 Denotation and covering .. 81
7.1.1 SitSpec templates . : 82
7.1.2 Covering .. : 85
7.1.3 Aktionsart .. : 86
7.2 Partial SemSpecs .. 87
7.2.1 Lexico-semantic combinations .. 87
7.2.2 Type shifting .. 89
7.2.3 Valency and the Upper Model .. 90
7.3 Alternations and extensions . 93
7.3.1 Alternations as meaning extensions . : 93
7.3.2 Lexical rules for alternations and extensions.. 95
7.3.3 Extension rules for circumstances . 101
7.3.4 Examples: lexical entries for verbs . : 102
7.3.5 Summary .. 104
7.4 Salience . . 105
7.5 Connotation .. 107
7.6 Summary: lexicalization with constraints and preferences .. 110
8 A new system architecture for multilingual generation 113
8.1 The computational problem . 113
8.2 Overview of the architecture . 115
8.2.1 Find lexical options . 115
8.2.2 Construct alternations and extensions .. 118
8.2.3 Establish preference ranking of options .. 119
8.2.4 Determine the complete and preferred SemSpec. 120
8.2.5 Generate sentence . : 122
8.3 Implementation of a prototype: MOOSE . 123
8.4 Embedding MOOSE in larger applications . : 125
9 Generating paraphrases 127
9.1 Verbalizing states .. 127
9.1.1 Binary states .. 127
9.1.2 Ternary states .. 128
9.2 Verbalizing activities .. 130
9.3 Verbalizing events .. : 132
9.4 Solutions to lexicalization problems. 138
10 Summary and conclusions 141
10.1 Summary of the work .. 141
10.2 Comparison to related work . 144
10.2.1 The role of the lexicon in NLG . : 144
10.2.2 Word{concept linking.. 144
10.2.3 Fine-grained lexical choices. 146
10.2.4 Paraphrasing .. 146
10.2.5 Event verbalization . : 148
10.2.6 Multilinguality and the lexicon . : 149
10.3 Contributions of the thesis . : 151
10.3.1 Lexical semantics for NLG. : 151
10.3.2 System architecture for NLG .. 152
10.3.3 Implementation .. 153
10.4 Directions for future research.. 153
Bibliography 156

List of Figures
1.1 Example of SitSpec: Jill filling a tank with water.. 6
1.2 Examples of SemSpecs and corresponding English sentences.. : 6
2.1 Lexicalization with 'zoom schemata' (from [Horacek 1990b]).. : 19
2.2 Small excerpt from Upper Model .. : : : 22
3.1 Taxonomy of eventualities from Bach [1986].. 35
5.1 Sample text from a Honda car manual .. 52
5.2 The top level of our ontology .. .. 55
5.3 Our classification of situation types .. : : 56
5.4 Event representation for Jill opening a wine bottle .. 57
5.5 LOOM definitions for basic ontological categories.. 59
5.6 Taxonomy of states .... : 61
5.7 LOOM definitions of binary-states .. : 62
5.8 LOOM definition of location-state .. 63
5.9 Subsumption of concepts and relations for ternary-states.. : 64
5.10 LOOM definition of path .... 65
5.11 Opening the wine bottle as transition.. : : 66
6.1 Representation levels in the generation system.. : 72
6.2 Syntax of SitSpecs .... 76
6.3 Example of situation specification as graph.. 77
6.4 Syntax of SemSpecs .... : 78
6.5 Semantic specifications and corresponding sentences .. 79
7.1 Syntax of a lexeme denotation .. .. 84
7.2 Syntax of partial SemSpecs .. .. 88
7.3 Example for type shifting .... 89
7.4 SitSpecs for sentences corresponding to configurations of to spray.. 98
7.5 Dependency of extension rules .. .. 100
7.6 Derivation of drain-configurations by extension rules .. 101
7.7 Sample lexical entries (abridged) for verbs.. : 103
8.1 Overall system architecture .. .. 116
8.2 Lexicon entries matching the SitSpec in fill{example, and their instantiations : : 118
8.3 Extension rules for fill{example, and resulting vos.. 119
8.4 The procedure for building SemSpecs (simplified).. 121
8.5 Screendump of Moose .... 124
9.1 SitSpec for water dripping from tank .. : 130
9.2 SitSpec for water rising in a tank .. : : : 132
9.3 SitSpec for Tom disconnecting the wire.. : : 134
9.4 SitSpec for Jill uncorking the bottle .. : 136
10.1 Lexicon entry for to require from ADVISOR II.. : 147
10.2 Sample CLCS and lexicon entries (abridged) from [Dorr 1993, pp. 224, 227] : : : 149