System and method for automatically generating translation templates from a pair of bilingual sentences

• US 5,442,546 A
• Filed: 11/30/1992
• Issued: 08/15/1995
• Est. Priority Date: 11/29/1991
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Morimoto, Yasutsugu, Kida, Yuko, Kaji, Hiroyuki
• Primary Examiner(s): Hayes, Gail O.
• Assistant Examiner(s): BODENDORF, ANDREW
• Application Number:
  US07/983,147
• Time in Patent Office: 988 Days
• Field of Search: 364/419.1, 364/419.07, 364/419.08, 364/419.14
• US Class Current: 704/4
• CPC Class Codes: G06F 40/45 Example-based machine translation; Alignment

Abstract

To automatically generate translation templates containing variables which can be replaced with various words or phrases from a bilingual pair of sentences, the machine translation system reads the first language sentence and second language sentence which are mutually equivalent, analyzes the morphemes and phrases of the sentences, identifies the word correspondence between the first language sentence and the second language sentence with reference to the bilingual dictionary, generates a translation template by replacing the corresponding words of the first language sentence and second language sentence with variables which are mutually correspondent, extracts the phrase correspondence between the first language sentence and the second language sentence, generates a generalized template wherein the corresponding phrases are replaced with variables, and generates a partial template wherein the corresponding phrases are separated. By doing this, a translation template can be learned (automatically generated) from bilingual pair of sentences, and high quality translation can be obtained.

Claims

1. A method of automatically generating a translation template in a system for automatically processing sentences in at least a first language and a second language, said method comprising the steps of:

reading a sentence in said first language and in said second language;

identifying correspondence between words of said first language sentence and said second language sentence which are read in said sentence reading step by referring to a bilingual dictionary between said first and second languages; and

generating a pair of a first language template and a second language template by selecting an optional subset from the set of word correspondences identified in said word correspondence identification step and replacing a corresponding word in both said first language sentence and said second language sentence with an identical variable for each correspondence belonging to the above subset,

said translation templates include a pair of templates of a sentence in said first language and a sentence in said second language which are respective translations of a same sentence, both of said pair of templates include a sentence including at least one variable which can be replaced with at least one of phrases or words.

2. A translation template generating method according to claim 1, said method further comprising the step of:

displaying the pair of said first language sentence and said second language sentence, which are the source of said generated translation template, in a format that parts which are replaced with variables and other parts are distinguished from each other.

3. A method of automatically generating a translation template according to claim 1, said method further comprising the steps of:

extracting candidate phrases from said first language sentence included in said translation template;

extracting candidate phrases from said second language sentence included in said translation template; and

generating a new pair of a first language template and a second language template by replacing both the candidate phrase in the first language template and the candidate phrase in the second language template with an identical variable when a pair of a candidate phrase extracted by said step of extracting candidate phrases from said first language sentence and a candidate phrase extracted by said step of extracting candidate phrases from said second language sentence which contain the same variable set is found.

4. A translation template generating method according to claim 3, said method further comprising the step of:

displaying the pair of said first language sentence and said second language sentence, which are the source of said generated translation template, in a format that parts which are replaced with variables and other parts are distinguished from each other.

5. A method of automatically generating a translation template according to claim 1, said method further comprising the steps of:

extracting candidate phrases from said first language sentence included in said translation template;

extracting candidate phrases from said second language sentence included in said translation template; and

separating the candidate phrase in said first language sentence and the candidate phrase in said second language sentence and generating a new translation template from the separated parts when a pair of a candidate phrase extracted by said step of extracting candidate phrases from said first language sentence and a candidate phrase extracted by said step of extracting candidate phrases from said second language sentence which contain the same variable set is found.

6. A translation template generating method according to claim 3, said method further comprising the step of:

displaying the pair of said first language sentence and said second language sentence, which are the source of said generated translation template, in a format that parts which are replaced with variables and other parts are distinguished from each other.

8. A translation template generating method according to claim 4, further comprising the step of:

judging whether said designated parts of said first language sentence are actual phrases.

7. A method of automatically generating a translation template in a system for automatically processing sentences in at least a first language and a second language, said method comprising the steps of:

reading a sentence in said first language and in said second language;

identifying correspondence between words of said first language sentence and said second language sentence which are read in said sentence reading step;

designating parts of said first language sentence to be replaced with variables;

extracting phrases of said second language sentence which correspond to the phrases of said designated parts of said first language sentence; and

generating a translation template by replacing the phrases of said designated parts of said first language sentence and the corresponding phrases of said second language sentence with variables,

said translation template includes a pair of templates of a sentence in said first language and a sentence in said second language which are respective translations of a same sentence, both of said pair of templates including a sentence including at least one variable which can be replaced with at least one of phrases or words.

9. A translation template generating method according to claim 7, said method further comprising the step of:

displaying results of said identifying step before said designating step is executed.

10. A translation template generating method according to claim 7, said method further comprising the step of:

displaying the pair of said first language sentence and said second language sentence, which are the source of said generated translation template, in a format that the parts which are replaced with variables and the other parts are distinguished from each other.

11. A method of automatically generating a translation template in a system for automatically processing sentences in at least a first language and a second language, said method comprising the steps of:

reading a sentence in said first language and in said second language;

identifying correspondence between the words of said first language sentence and said second language sentence which are read in said bilingual pair of sentence reading step;

designating parts of said first language sentence and said second language sentence to be replaced with variables; and

generating a translation template by replacing said designated parts of said first language sentence and said designated parts of said second language sentence with variables,

said translation template includes a pair of templates of a sentence in said first language and a sentence in said second language which are respective translations of a same sentence, both of said pair of templates including a sentence including at least one variable which can be replaced with at least one of phrases or words.

12. A translation template generating method according to claim 11, said method further comprising the steps of:

judging whether said designated parts of said first language sentence are actually phrases;

judging whether said designated parts of said second language sentence are actually phrases; and

judging whether said designated parts of said first language sentence and said designated parts of said second language sentence are mutually correspondent.

13. A translation template generating method according to claim 11, said method further comprising the step of:

displaying results of said identifying step before said designating step is executed.

14. A translation template generating method according to claim 11, said method further comprising the step of:

displaying the pair of said first language sentence and said second language sentence, which are the source of said generated translation template, in a format that parts which are replaced with variables and other parts are distinguished from each other.

15. A system for automatically generating a translation template by processing sentences in at least a first language and a second language, comprising:

bilingual sentence reading means for reading a sentence in said first language and in said second language;

identification means for identifying correspondence between words of said first language sentence and said second language sentence read by said bilingual sentence reading means by referring to a bilingual dictionary between said first and second languages; and

translation template generation means for generating a pair of a first language template and a second language template by selecting an optional subset from said correspondence identified by said identification means and replacing both words in said first language sentence and words in said second language sentence with identical variables for each identified correspondence,

said translation template including a sentence in said first language and a sentence in said second language which are respective translations of a same sentence, both of said templates including a sentence including at least one variable which can be replaced with at least one of phrases or words.

16. A translation template learning system according to claim 15, further comprising:

means for designating parts of said first language sentence to be replaced with variables.

17. A translation template learning system according to claim 15, further comprising:

means for designating the parts of said first language sentence and said second language sentence to be replaced with variables.

18. A translation template learning system according to claim 15, further comprising:

means for displaying said word correspondence between pair of sentences identification result.

19. A translation template learning system according to claim 15, further comprising:

means for displaying the pair of first language sentence and second language sentence, which are the source of the generated translation template, in a format that parts which are replaced with variables and other parts are distinguished from each other.