Statistical machine translation

• US 7,624,005 B2
• Filed: 03/28/2003
• Issued: 11/24/2009
• Est. Priority Date: 03/28/2002
• Status: Active Grant

• Current Assignee: University of Southern California
• Original Assignee: University of Southern California
• Inventors: Koehn, Philipp, Knight, Kevin
• Primary Examiner(s): Dorvil; Richemond
• Assistant Examiner(s): BAKER, MATTHEW H
• Application Number:
  US10/403,862
• Publication Number:
  US 20040024581A1
• Time in Patent Office: 2,433 Days
• Field of Search: 704/4
• US Class Current: 704/2
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/44 Statistical methods, e.g. probability models
• CPC Class Codes: G06F 40/45 Example-based machine translation; Alignment

Abstract

A method includes detecting a syntactic chunk in a source string in a first language, assigning a syntactic label to the detected syntactic chunk in the source string, mapping the detected syntactic chunk in the source string to a syntactic chunk in a target string in a second language, said mapping based on the assigned syntactic label, and translating the source string into a possible translation in the second language.

Claims

1. A computer implemented method, the method comprising:

assigning a part of speech identifier to each word in a source string, the source string in a first language;

detecting a first sequence of syntactic chunks in the source string, the syntactic chunks each comprising at least one of the words;

assigning a syntactic chunk label to each of the detected syntactic chunks in the source string;

defining connections between each of the detected syntactic chunks in the source string and at least one syntactic chunk of a sequence of syntactic chunks in a target string, the target string being a parallel translation in a second language of the source string, said defining comprising determining connections based on a chunk mapping table, the chunk mapping table using pre-defined connections based on the assigned syntactic chunk label;

mapping each word in the detected syntactic chunks in the source string to each word in the syntactic chunks in the target string, said mapping based on a word mapping table and the part of speech identifier;

translating by a computer an input string in the first language into a translation in the second language based on the chunk mapping table and the word mapping table, and

outputting the translation using the computer.

2. The method of claim 1, wherein assigning the syntactic chunk label comprises assigning based on the assigned part of speech identifier of the at least one of the words in the source string.

3. The method of claim 1, wherein defining connections comprises defining a connection between the detected chunk from the source string to at least two non-adjacent chunks in the target string.

4. The method of claim 1, wherein defining connections comprises defining a connection of at least two detected chunks from the source string to a single chunk in the target string.

5. The method of claim 1, wherein translating comprises including at least two words in the target string that corresponds to a single word in the source string.

6. The method of claim 1, wherein translating comprises translating phrases.

7. The method of claim 1, further comprising:

applying a language model to the source string, the language model based upon the language of the target string.

8. The method of claim 1, further comprising:

determining a probability of said mapping.

9. The method of claim 1, wherein translating comprises inserting at least one NULL word in the target string.

10. A computer readable storage medium having stored thereon a program, the program being executable by a processor for performing a method, the method comprising:

assigning a part of speech identifier to each word in a source string, the source string in a first language;

detecting a first sequence of syntactic chunks in the source string, the syntactic chunks each comprising at least one of the words;

assigning a syntactic chunk label to each of the detected syntactic chunks in the source string;

defining connections between each of the detected syntactic chunks in the source string and at least one syntactic chunk of a sequence of syntactic chunks in a target string, the target string being a parallel translation in a second language of the source string, said defining comprising determining connections based on a chunk mapping table, the chunk mapping table using pre-defined connections based on the assigned syntactic chunk label;

mapping each word in the detected syntactic chunks in the source string to each word in the syntactic chunks in the target string, said mapping based on a word mapping table and the part of speech identifier; and

translating by a computer an input string in the first language into a translation in the second language based on the chunk mapping table and the word mapping table.

11. The computer readable medium of claim 10, wherein assigning the syntactic chunk label further comprises assigning the syntactic chunk label based on the assigned part of speech identifier of the at least one of the words in the source string.

12. The computer readable medium of claim 10, wherein defining connections comprises defining a connection between the detected chunk from the source string to at least two non-adjacent chunks in the target string.

13. The computer readable medium of claim 10, wherein defining connections comprises defining a connection of at least two detected chunks from the source string to a single chunk in the target string.

14. The computer readable medium of claim 10, wherein translating comprises including at least two words in the target string that corresponds to a single word in the source string.

15. The computer readable medium of claim 10, wherein translating comprises causing a machine to translate phrases.

16. The computer readable medium of claim 10, wherein the method further comprises:

applying a language model to the source string, the language model based upon the language of the target string.

17. The computer readable medium of claim 10, wherein the method further comprises:

determining a probability of said mapping.

18. The computer readable medium of claim 10, wherein translating comprises causing a machine to insert at least one NULL word in the target string.