汉语篇章零形回档的解析与生成

前言

ABBREVIATIONS

TABLES AND FIGURES

Chapter 1 INTRODUCTION

1.1 What is anaphora？

1.2 Types of anaphora

1.3 Anaphora resolution and generation

1.4 Scope and goal of the study

1.5 Organization of the study

Chapter 2 PREVIOUS APPROACHES TO ZERO ANAPHORA IN CHINESE

2.1 Introduction

2.2 A syntactic approach

2.3 An extra-syntactic approach

2.3.1 The discourse framework

2.3.1.1 Li and Thompson‘s conjoinabihty constraint

2.3.1.2 Chen’s predictability condition and neg- ligibility condition

2.3.1.3 Zhou‘s global coreference

2.3.1.4 Xu’s functional approach

2.3.1.5 Tao‘s emergent reference

2.3.1.6 Cheng and Lee’s recovery principles

2.3.1.7 You‘s recovery rules

2.3.1.8 Xu’s resolution principle

2.3.2 Huang‘s pragmatic approach

2.3.3 Tomlin and Pu’s cognitive approach

2.4 Summary

Chapter 3 CENTERING THEORY

3.1 Introduction

3.2 Centering theory

3.2.1 General description

3.2.2 Coherence and attentional state

3.2.3 Coherence and referring expression form

3.2.4 Centering definitions and constraints

3.3 Summary

Chapter 4 RESOLVING CHINESE ZERO ANA- PHORS WITH RICM

4.1 Introduction

4.2 BFP algorithm for centering and anaphora resolution

4.3 Implementation of BFP

4.4 Problems with Centering and BFP

4.5 Previous proposals for possible solution

4.5.1 The Stack Model

4.5.2 The Global Model

4.5.3 The Cache Model

4.5.3.1 Basic notions of the Cache Model

4.5.3.2 Integrating cache model with centering algorithm

4.5.3.3 Implementation of ICM

4.5.3.4 Evaluation of ICM

4.6 A Revised Cache Model

4.6.1 Introduction

4.6.2 Theoretical assumptions

4.6.3 Revision of Centering Rule 1（Pronoun Rule）

4.6.4 Integrating the revised cache model with centering algorithm

4.6.5 Implementation of RICM

4.6.6 A simplified version of RICM

4.6.7 Evaluation of RICM

4.6.8 Experiment

4.6.8.1 Data

4.6.8.2 Procedure

4.6.8.3 Result

4.6.8.4 Discussion

4.7 Summary

Chapter 5 SETTING PARAMETERS FOR CEN- TERING ANALYSIS ON CHINESE DISCOURSE

5.1 Introduction

5.2 Specifying the utterance unit

5.2.1 Introduction

5.2.2 What is an utterance？

5.2.3 Utterance for Centering analysis

5.2.4 Intermediate summary

5.3 Discourse segmentation

5.3.1 The difficulty with discourse segmentation

5.3.2 Problems arising from different segmentations

5.3.3 Specifying discourse segment for Chinese

5.3.3.1 Topic chain as discourse segment

5.3.3.2 Topic continuity as discourse segment

5.4 Ranking forward-looking centers in Chinese

5.4.1 Related cross-linguistic work

5.4.2 Topic as a grammatical concept

5.4.3 Topic and reference

5.4.4 The ranking hierarchy

5.4.5 Other factors contributing to the salience of entities

5.4.6 Ranking complex NPs

5.5 Summary

Chapter 6 GENERATING CHINESE ZERO ANAPHORS

6.1 Introduction

6.2 Previous Centering approaches to generating （zero） anaphors

6.3 Assumptions

6.4 Corpus study

6.4.1 Corpus and coding

6.4.2 Results and analysis

6.5 Developing zero generation algorithm

6.5.1 Zero generation algorithm（I）

6.5.2 The privilege of transition pairs

6.5.3 Zero generation algorithm（l）

6.5.4 Zero generation algorithm（m）

6.6 Summary

Chapter 7 COMPUTATION OF TRANSITION TYPES

7.1 Introduction

7.2 Transitions for modeling coherence

7.3 The problem of inferable centers

7.4 Inferable centers as bridging references

7.5 Possible solutions to inferable centers

7.5.1 Laurel Fals‘s proposal

7.5.2 Extension of LFP

7.5.2.1 Overestimation of complete shifts

7.5.2.2 The “Cheapness” Principle by SHP

7.5.2.3 Integrating LFP with SHP

7.6 Summary

Chapter 8 CONCLUSION

8.1 Major findings

8.2 Limitations

8.3 Directions for future study

REFERENCES

《汉语篇章零形回指的解析与生成》的特色是将当前国外先进的计算语言学理论“向心理论”有效应用于汉语篇章零形回指的解析与生成，提出的模型和算法经过实际语料的验证可有效解析和生成汉语篇章零形回指，它对于对外汉语教学中回指的确认和理解具有一定的启发和辅助作用，尤其对于包括机器翻译、自动文本概要和信息提取在内的自然语言处理具有一定的应用价值。篇章零形回指的解析与生成是汉语篇章处理的关键环节，也是当前篇章处理研究的热点之一，而就目前的国内外研究现状来看，绝大部分研究都缺乏明晰的表征手段和形式化操作模式。