Gao Wei. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2004. / Includes bibliographical references (leaves 79-82). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iii / Bibliographic Notes --- p.v / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- What is Transliteration? --- p.1 / Chapter 1.2 --- Existing Problems --- p.2 / Chapter 1.3 --- Objectives --- p.4 / Chapter 1.4 --- Outline --- p.4 / Chapter 2 --- Background --- p.6 / Chapter 2.1 --- Source-channel Model --- p.6 / Chapter 2.2 --- Transliteration for English-Chinese --- p.8 / Chapter 2.2.1 --- Rule-based Approach --- p.8 / Chapter 2.2.2 --- Similarity-based Framework --- p.8 / Chapter 2.2.3 --- Direct Semi-Statistical Approach --- p.9 / Chapter 2.2.4 --- Source-channel-based Approach --- p.11 / Chapter 2.3 --- Chapter Summary --- p.14 / Chapter 3 --- Transliteration Baseline --- p.15 / Chapter 3.1 --- Transliteration Using IBM SMT --- p.15 / Chapter 3.1.1 --- Introduction --- p.15 / Chapter 3.1.2 --- GIZA++ for Transliteration Modeling --- p.16 / Chapter 3.1.3 --- CMU-Cambridge Toolkits for Language Modeling --- p.21 / Chapter 3.1.4 --- Re Write Decoder for Decoding --- p.21 / Chapter 3.2 --- Limitations of IBM SMT --- p.22 / Chapter 3.3 --- Experiments Using IBM SMT --- p.25 / Chapter 3.3.1 --- Data Preparation --- p.25 / Chapter 3.3.2 --- Performance Measurement --- p.27 / Chapter 3.3.3 --- Experimental Results --- p.27 / Chapter 3.4 --- Chapter Summary --- p.28 / Chapter 4 --- Direct Transliteration Modeling --- p.29 / Chapter 4.1 --- Soundness of the Direct Model一Direct-1 --- p.30 / Chapter 4.2 --- Alignment of Phoneme Chunks --- p.31 / Chapter 4.3 --- Transliteration Model Training --- p.33 / Chapter 4.3.1 --- EM Training for Symbol-mappings --- p.33 / Chapter 4.3.2 --- WFST for Phonetic Transition --- p.36 / Chapter 4.3.3 --- Issues for Incorrect Syllables --- p.36 / Chapter 4.4 --- Language Model Training --- p.36 / Chapter 4.5 --- Search Algorithm --- p.39 / Chapter 4.6 --- Experimental Results --- p.41 / Chapter 4.6.1 --- Experiment I: C.A. Distribution --- p.41 / Chapter 4.6.2 --- Experiment II: Top-n Accuracy --- p.41 / Chapter 4.6.3 --- Experiment III: Comparisons with the Baseline --- p.43 / Chapter 4.6.4 --- Experiment IV： Influence of m Candidates --- p.43 / Chapter 4.7 --- Discussions --- p.43 / Chapter 4.8 --- Chapter Summary --- p.46 / Chapter 5 --- Improving Direct Transliteration --- p.47 / Chapter 5.1 --- Improved Direct Model´ؤDirect-2 --- p.47 / Chapter 5.1.1 --- Enlightenment from Source-Channel --- p.47 / Chapter 5.1.2 --- Using Contextual Features --- p.48 / Chapter 5.1.3 --- Estimation Based on MaxEnt --- p.49 / Chapter 5.1.4 --- Features for Transliteration --- p.51 / Chapter 5.2 --- Direct-2 Model Training --- p.53 / Chapter 5.2.1 --- Procedure and Results --- p.53 / Chapter 5.2.2 --- Discussions --- p.53 / Chapter 5.3 --- Refining the Model Direct-2 --- p.55 / Chapter 5.3.1 --- Refinement Solutions --- p.55 / Chapter 5.3.2 --- Direct-2R Model Training --- p.56 / Chapter 5.4 --- Evaluation --- p.57 / Chapter 5.4.1 --- Search Algorithm --- p.57 / Chapter 5.4.2 --- Direct Transliteration Models vs. Baseline --- p.59 / Chapter 5.4.3 --- Direct-2 vs. Direct-2R --- p.63 / Chapter 5.4.4 --- Experiments on Direct-2R --- p.65 / Chapter 5.5 --- Chapter Summary --- p.71 / Chapter 6 --- Conclusions --- p.72 / Chapter 6.1 --- Thesis Summary --- p.72 / Chapter 6.2 --- Cross Language Applications --- p.73 / Chapter 6.3 --- Future Work and Directions --- p.74 / Chapter A --- IPA-ARPABET Symbol Mapping Table --- p.77 / Bibliography --- p.82
| Identifer | oai:union.ndltd.org:cuhk.edu.hk/oai:cuhk-dr:cuhk_324956 |
| Date | January 2004 |
| Contributors | Gao, Wei., Chinese University of Hong Kong Graduate School. Division of Systems Engineering and Engineering Management. |
| Source Sets | The Chinese University of Hong Kong |
| Language | English, Chinese |
| Detected Language | English |
| Type | Text, bibliography |
| Format | print, xii, 82 leaves : ill. ; 30 cm. |
| Rights | Use of this resource is governed by the terms and conditions of the Creative Commons “Attribution-NonCommercial-NoDerivatives 4.0 International” License (http://creativecommons.org/licenses/by-nc-nd/4.0/) |
Page generated in 0.0073 seconds