This book introduces an automatic grammatical error correction model on sentence-level based on deep learning. For the first time in error correction tasks, native Chinese corpus (literary works) and sentence classification information (acceptable or unacceptable) were incorporated into the Transformer model for training. Previous research often overlooked the significant impact of native language information on error correction. Experiments demonstrated a substantial improvement in model accuracy, increasing from 9.7% to 48% with the inclusion of native corpus and classification information.
Sample Chapter(s)
Preface (166 KB)
Components of the Book:
- Preface
- About the Author
- Figure Index
- Table Index
- Chapter 1: Introduction
- 1.1 Background and Significance of the Study
- 1.2 Significance of the Study
- 1.3 Literature Review
- 1.4 Research Objectives
- 1.5 Research Methods
- Chapter 2: Error Analysis and Knowledge Base Construction for Chinese as a Second Language Learning
- 2.1 Introduction
- 2.2 Analysis of Chinese Sentence Errors Based on the HSK Dynamic
Composition Corpus
- 2.3 Construction of a Linguistic Dictionary for Correcting Errors in
Chinese Sentences
- 2.4 Construction of Sentence Acceptability Rating Test Corpus
- 2.5 Summary
- Chapter 3: Research on a Chinese Knowledge Representation Model for Error Analysis
- 3.1 Introduction
- 3.2 A Hierarchical Context and Order Based Encoding Model
- 3.3 Integrating Character Embeddings and Character Order into the
Chinese Memory Vector Model
- 3.4 Language Model Integrating Pre-Trained Character Embeddings
- 3.5 Experiments and Results Analysis
- 3.6 Summary
- Chapter 4: Research on Algorithms for Chinese Sentence Acceptability in Error Analysis
- 4.1 Introduction
- 4.2 Methods for Converting Sentence Acceptability
- 4.3 Evaluation Metrics
- 4.4 Exploration of Unsupervised Sentence Acceptability Algorithms
- 4.5 Chinese Sentence Acceptability for Error Analysis
- 4.6 Exploring the Relationship between Language Model Perplexity and
Sentence Acceptability
- 4.7 Summary
- Chapter 5: Research on a Chinese Grammatical Error Correction Model on Sentence-Level Based on Deep Learning
- 5.1 Introduction
- 5.2 Deep Learning-Based Error Correction Model on Sentence-Level
- 5.3 Error Correction on Character and Word-Level Based on Error Information
- 5.4 Ranking Method Based on Sentence Acceptability
- 5.5 Experiments and Results Analysis
- 5.6 Summary
- Chapter 6: Summary and Future Prospects
- 6.1 Contributions
- 6.2 Limitations
- 6.3 Future Research Directions
- Works Cited
Readership:
Students, academics, teachers and other people attending or interested in the situation of the Nubi Community in Eastern Africa.
|
1
|
|
Preface
WANG, Jing
PDF (166 KB)
|
|
|
|
|
1
|
|
About the Author
WANG, Jing
PDF (110 KB)
|
|
|
|
|
1
|
|
Figure Index
WANG, Jing
PDF (196 KB)
|
|
|
|
|
1
|
|
Table Index
WANG, Jing
PDF (197 KB)
|
|
|
|
|
1
|
|
Chapter 1: Introduction
WANG, Jing
PDF (1577 KB)
|
|
|
|
|
51
|
|
Chapter 2: Error Analysis and Knowledge Base Construction for Chinese as a Second Language Learning
WANG, Jing
PDF (775 KB)
|
|
|
|
|
85
|
|
Chapter 3: Research on a Chinese Knowledge Representation Model for Error Analysis
WANG, Jing
PDF (844 KB)
|
|
|
|
|
117
|
|
Chapter 4: Research on Algorithms for Chinese Sentence Acceptability in Error Analysis
WANG, Jing
PDF (805 KB)
|
|
|
|
|
141
|
|
Chapter 5: Research on a Chinese Grammatical Error Correction Model on Sentence-Level Based on Deep Learning
WANG, Jing
PDF (1994 KB)
|
|
|
|
|
171
|
|
Chapter 6: Summary and Future Prospects
WANG, Jing
PDF (212 KB)
|
|
|
|
|
177
|
|
Works Cited
WANG, Jing
PDF (329 KB)
|
|
|
|
Jing WANG
Artificial Intelligence and Language Cognition Laboratory, Bei-jing International Studies University, Beijing, China, 100024.