Method and apparatus for example-based machine translation with learned word associations

• US 20040172235A1
• Filed: 02/28/2003
• Published: 09/02/2004
• Est. Priority Date: 02/28/2003
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Smets, Martine, Pinkham, Jessie
• Granted Patent: US 7,356,457 B2
• Time in Patent Office: Days
• US Class Current: 704/2
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars

Abstract

A method and computer-readable medium are provided that perform a series of steps associated with machine translation. These steps include using a first text in a first language and a second text in a second language, to produce an association list where words in the first language are associated with words in the second language. A first syntactic structure for a sentence from the first text is aligned with a second syntactic structure for a sentence in the second text based on the association list without referring to a bilingual dictionary of content words. The association list is also used during translations. Specifically, a word in the first language is translated into a word in the second language based on an entry in the association list without referring to a bilingual dictionary that contains content words. Thus, training and translation are performed without the need for a bilingual dictionary of content words.

Claims

1. A method of translating words in a first language into words in a second language, the method comprising:

using a first training text in a first language and a second training text in a second language to generate a word association list comprising entries, at least one of the entries comprising a word from the first language and a word from the second language; and

translating words in the first language into words in the second language based in part on the word association list without referring to a multi-lingual human-authored dictionary of content words wherein a content word comprises a word that is not a preposition, is not a conjunction, is not a pronoun and is not an article.

2. The method of claim 1 further comprising using the word association list to perform an alignment of an element of a sentence in the first training text with an element of a sentence in the second training text.

3. The method of claim 1 wherein performing an alignment comprises not referring to a multi-lingual human-authored dictionary of content words.

4. The method of claim 3 further comprising generating a first logical form for a sentence in the first training text and generating a second logical form for a sentence in the second training text wherein aligning an element of a sentence in the first training text with an element of a sentence in the second training text comprises aligning the first logical form with the second logical form.

5. The method of claim 1 wherein translating a word in the first language into a word in the second language comprises determining a part of speech of a word in the first language and searching for the word and the part of speech in the word association list.

6. The method of claim 5 wherein searching for the word and the part of speech further comprises searching for a part of speech for a word in the second language.

7. The method of claim 6 wherein the part of speech of the word in the first language matches the part of speech of a word in the second language.

8. A computer-readable medium having computer-executable instructions for performing steps comprising:

identifying associations between words in a first language and words in a second language based on a first text in the first language and a second text in the second language to produce an association list;

identifying a first syntactic structure for a sentence in a first language;

identifying a second syntactic structure for a sentence in a second language; and

aligning the first syntactic structure with the second syntactic structure based on the association list without referring to a human-authored bilingual dictionary of content words.

9. The computer-readable medium of claim 8 wherein the association list comprises a set of entries, at least one of the entries comprising a single word from the first language and a single word from the second language.

10. The computer-readable medium of claim 9 wherein the computer-executable instructions perform further steps comprising translating a sentence in the first language into a sentence in the second language based on the association list.

11. The computer-readable medium of claim 10 wherein translating comprises not referring to a human-authored bilingual dictionary of content words.

12. The computer-readable medium of claim 10 wherein at least one entry in the association list comprises a single word from the first language, a part of speech for the single word from the first language, a single word from the second language and a part of speech for the single word from the second language.

13. The computer-readable medium of claim 12 wherein translating a sentence in the first language into a sentence in the second language comprises searching for an entry in the association list that contains a word from the sentence in the first language.

14. The computer-readable medium of claim 13 wherein searching for an entry further comprises searching for an entry that contains the part of speech for the word from the sentence in the first language.

15. The computer-readable medium of claim 14 wherein searching for an entry further comprises searching for an entry that contains a particular part of speech for a word from the second language.

16. The computer-readable medium of claim 15 wherein the particular part of speech for a word from the second language is the same as the part of speech for the word from the sentence in the first language.