Dissertation > Excellent graduate degree dissertation topics show

A Comparative Study of Rule-based and Statistical-based Machine Translation Models

Author: WangZuo
Tutor: WuZhongJie
School: University of Electronic Science and Technology
Course: Foreign Linguistics and Applied Linguistics
Keywords: MT models automatic evaluation Google Translate SYSTRANet error analysis
CLC: H085
Type: Master's thesis
Year: 2011
Downloads: 67
Quote: 0
Read: Download Dissertation

Abstract


This thesis mainly discusses the working process, architectural structures, as well as the advantages and disadvantages of mainstream machine translation models. It is found that, in previous researches, most of them focused on the comparison of MT product qualities in the text level or in their commercial or economical values, while few researches have explored the working systems of MT models which determine, to a large degree, the accuracy of translation product. In this study, three research questions are addressed:1) how the mechanics function in the translating process of each model; 2) how linguistic features are dealt with in the process of machine translation; 3) What improvements can be suggested on the accuracy and fluency of machine translation.For a quantitative study, seventy-five pieces of bilingual news reports covering various topics are collected from China Daily’s official online website as the sample texts for evaluation. All the sample Chinese texts are processed by two MT systems, Google Translate and SYSTRANet Translation, for their English versions. Then, an automatic evaluation of the MT products is executed through BLEU metric for an overall investigation of the RBMT model and SBMT model. It is revealed that Google Translate is superior to SYSTRANet in accuracy of translation on vocabulary level.Human evaluation is carried out by marking and sorting translation errors frequently found in the translation by the two MT systems. The findings demonstrate that the two MT models differ in dealing with different aspects of linguistic features in their translation. While Google Translate has advantages in processing lexical and semantic meaning of the source texts, SYSTRANet keeps better word orders and syntactical relations.In addition, typical translation errors of the two MT systems are used as examples for a qualitative analysis in the thesis. It is revealed that Google Translate can process comparative structures better in the lexical level, while SYSTRANet can work better on complex syntactic structures such as attributive clauses.The fruits of the research provide a reliable analysis of the performances of different MT models on the same sample texts. It has turned out that each MT model has both advantages and disadvantages, as nothing like an overall superior MT model has been found. Based on the findings, suggestions on improving MT are presented for future research.

Related Dissertations

  1. Error Analysis and Measurement Method Design for the Hemispherical Resonator Gyro,V241.5
  2. Korean Students ' Acquisition of Chinese Classifiers error analysis,H195
  3. English -speaking Chinese learners willing verb bias analysis,H195
  4. South Korean students learning Chinese in the error analysis,H195
  5. The Error Analysis and Teaching Strategy of "Ba" Sentences to Foreign Students,H195
  6. Error Analysis of Prepositions and Its Teaching Strategy for Primary Stage of Korean Learners in Learning Chinese,H195
  7. The Study of the Chinese Sentence Structure "Shi……de" in the Chinese Language Teaching for Korean Students,H195
  8. Chinese Personal Affix (Quasi-affixes) in Teaching Chinese as a Foreign Language,H195
  9. Analysis on the Biased Errors of Word Order in Written Expression of Foreign Students,H195
  10. The Research of Detachable Words and Teaching Chinese as a Second Language,H195
  11. The Similarities and Differences among "Shi" "Ling" "Rang" Sentences in Modern Chinese,H146
  12. Error Analysis of English Translation of Chinese Public Signs,H315.9
  13. The Commparative Study of Chinese and Korea Quantifier,H14
  14. Elementary level of Thai students \,H195
  15. A Corpus-based Study on the Vocabulary Errors in the Writings of Chinese EFL Learners and Its Pedagogical Implications,H319
  16. The Pronunciation Error Analysis and Teaching Countermeasures to Thai Undergraduates of Chinese,H195
  17. Large spiral bevel gear error analysis and correction techniques,TH132.41
  18. The Analysis of Acquisition of Liheci in Chinese of Intermediate and Advanced Thai Students,H195
  19. Error Analysis of Thai Sudent in Using Adverb "You" and "Zai",H195
  20. Adverb \,H195
  21. Error Analysis on Degree Adverb "Hen" in Interlanguages and the Teaching Methods,H146

CLC: > Language, writing > Linguistics > Applied Linguistics > Machine Translation
© 2012 www.DissertationTopic.Net  Mobile