Research on Semantic-based Passive Transformation in Chinese-English Machine Translation Wenfei Chang, Zhiying Liu, Yaohong Jin Institute o

Research on Semantic-based Passive Transformation in Chinese-English Machine Translation  
 

Wenfei Chang, Zhiying Liu, Yaohong Jin

Institute of Chinese Information Processing

Beijing Normal University

Introduction 

1 

          Semantic analysis of passive voice 

2 

Transformation rules and algorithm 

3 

Experiments and Result Analysis 

4 

Conclusions 

5 

Outline

1. Introduction

Type  

Sentence number 

  Proportion 

Sentences with passive mark   

390 

39% 

Sentences without passive mark 

610 

61% 

2. Semantic analysis of passive voice 

     We have investigated 1000 sentences which should be transformed into English when translating.  

Table 1. Classification of Passive Sentence

• Sentences with passive mark in Chinese

  

  

• 因此提交订单的交易者将被通知成交。(Thereby  the  trader  that  sent in the order will be informed about the deal.)

• 它不需要处理在第一排列单元所接收的订单。(It does not need to handle the order that  was received at the first ranking unit.) 
   
   
  

 

2. Semantic analysis of passive voice 
 
 
 

Passive mark BEI 
 
 
 

Passive mark SUO

ALL_PASS 

passive voice will be used in English 

Verb+ Prep 

2. Semantic analysis of passive voice 

• Sentences without passive mark in Chinese

• For example,

 

  

      1、 “V+NP”

• 经由所述双向隧道转发分组。(Packets are forwarded via the bi-directional tunnel.)

     2、 “NP+V”

• 固定的和旋转的磁鼓面对面地安装。(The fixed and rotatable drums are installed face to face.) 

• 包套可滑动地安装在可弯曲管内。(A sheath is slideably mounted inside the flexible pipe .) 
   
  

  

• 这种组合物可以做成很薄、很小的产品。（The composition can be made into a very thin and small product.）

 
 
 
 

Component  ellipsis  in  sentence.  
 
 
 

 “Verb+Prep” structure in sentence.  
 
 
 

Effect Sentence 

2. Semantic analysis of passive voice

      A series of rules are drawn up according to several situations . The specific steps are as fellows: 
 

3. Transformation rules and algorithm

Type 
 

Total number 

Should be transformed 

Transformed 

Right transformed 

RB 
 

1000 
 

632 
 

540 
 

481 

Google 
 

1000 
 

632 
 

515 
 

430 
 

System 

Precision 

Recall 

RB 

89.1% 

76.1% 

Google 

83.4% 

68.1% 

Table 2. Types of data 

Table 3. Result of transformation 

4. Experiments and Result Analysis

• Rules have not covered all kinds of linguistic phenomenon.

• Knowledge base gives wrong property (“ALL_PASS[Y]”) to the verb. 
  

• The verb is wrongly recognized, thus leading to wrongly match the transformation rules.  
  

 

4. Experiments and Result Analysis 

By analyzing errors in the result, we find there are mainly have three reasons:

•     Results show that our system has achieved a good effect.

•     In the future, we will make further improvements based on the errors.   
  

 

5. Conclusions

 
 

    Thank    you   !