TRANSLATION OF MULTILINGUAL EMBEDDED PHRASES

• US 20120253785A1
• Filed: 04/03/2011
• Published: 10/04/2012
• Est. Priority Date: 04/03/2011
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hamid, Ahmed Abdul, Darwish, Kareem
• Granted Patent: US 9,098,488 B2
• Time in Patent Office: Days
• US Class Current: 704/4
• CPC Class Codes: G06F 40/131 Fragmentation of text files, e.g. creating reusable text-blocks; Linking to fragments, e.g. using XInclude; Namespaces
• CPC Class Codes: G06F 40/263 Language identification
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation

Abstract

A communication object including a plurality of object words may be received. The communication object may be parsed to identify each of the object words as tokens. A first natural language and at least one natural language different from the first natural language that are associated with the plurality of object words may be determined, based on a language analysis of the tokens. Tokens associated with the first natural language and tokens included in embedded word phrases associated with the embedded natural language may be translated, via a translating device processor, to a target natural language, based on at least one context associated with the communication object.

Claims

1. A system comprising:

a multilingual translator engine that includes;

a receiving engine configured to receive a communication object including a plurality of object words arranged in sequence;

a language determination engine configured to determine a first natural language and at least one embedded natural language different from the first natural language associated with the communication object, based on at least one first language word phrase associated with the first natural language and at least one embedded word phrase associated with the embedded natural language that are included in the plurality of object words; and

an embedded phrase translator engine configured to translate, via a translating device processor, the at least one embedded word phrase to a target natural language, based on at least one context associated with the communication object.

2. The system of claim 1, further comprising:

a transliterated word detection engine configured to determine at least one transliterated word included in the communication object, each transliterated word including a transliterated character string; and

a transliterated word conversion engine configured to convert the at least one transliterated word to one of the at least one embedded natural language, the first natural language, and the target natural language.

3. The system of claim 1, wherein:

the target natural language includes the first natural language.

4. The system of claim 1, further comprising:

a translation model probability engine configured to determine a translation model probability value associated with a candidate translational phrase, based on a frequency of occurrence value associated with the candidate translational phrase; and

a language model probability engine configured to determine a language model probability value associated with the candidate translational phrase, based on a predictive value associated with a sequential word context associated with a language model,

wherein the embedded phrase translator engine is configured to translate, via the translating device processor, the at least one embedded word phrase to the target natural language, based on combining the translation model probability value and the language model probability value that are associated with candidate translational phrases associated with the communication object.

5. The system of claim 4, wherein:

the embedded phrase translator engine is configured to translate, via the translating device processor, the at least one embedded word phrase to the target natural language, based on one of;

an arithmetic product of the translation model probability value and the language model probability value, and

an arithmetic sum of the log of the translation model probability value and log of the language model probability value.

6. The system of claim 4, further comprising:

a translation length determination engine configured to determine a length of a candidate translation of a group of the object words; and

a phrase length determination engine configured to determine a difference between a length of a phrase of object words associated with the at least one embedded natural language and a length of a phrase associated with the target natural language that corresponds to a candidate translation of the phrase of object words,

wherein the embedded phrase translator engine is configured to translate the at least one embedded word phrase to the target natural language, based on at least one of;

the length of the candidate translation of the group of the object words, and

the difference between the length of the phrase of object words associated with the at least one embedded natural language and the length of the phrase associated with the target natural language that corresponds to the candidate translation of the phrase of object words.

7. The system of claim 1, further comprising:

a parsing engine configured to recognize each of the plurality of object words as a token,

wherein the language determination engine is configured to identify a determined natural language associated with each word recognized by the parsing engine.

8. The system of claim 7, wherein:

the determined natural language includes one of the at least one embedded natural language and the first natural language.

9. The system of claim 7, wherein:

the language determination engine is configured to determine the first natural language and the at least one embedded natural language, based on a received user input specifying the first natural language and the at least one embedded natural language.

10. The system of claim 7, wherein:

the language determination engine is configured to determine the first natural language and the at least one embedded natural language, based on a language analysis of at least one of the object words recognized by the parsing engine.

11. A computer program product tangibly embodied on a computer-readable medium and including executable code that, when executed, is configured to cause at least one data processing apparatus to:

receive a communication object including a plurality of object words arranged in sequence;

parse the communication object to identify each of the object words as tokens;

identify at least one of the tokens as an embedded transliterated word that includes a transliterated character string associated with a first natural language;

identify at least another one of the tokens as a non-transliterated word; and

convert the embedded transliterated word to a corresponding non-transliterated character string representation based on a script representation associated with the first natural language.

12. The computer program product of claim 11, wherein the executable code, when executed, is configured to cause the at least one data processing apparatus to:

determine the first natural language and at least one embedded natural language different from the first natural language, that are associated with the plurality of object words; and

translate tokens associated with the first natural language and tokens included in embedded word phrases associated with the at least one embedded natural language to a target sequence of words associated with a target natural language, based on at least one context associated with the communication object.

13. The computer program product of claim 12, wherein the executable code, when executed, is configured to cause the at least one data processing apparatus to:

translate the tokens associated with the first natural language and the tokens included in embedded word phrases associated with the at least one embedded natural language to a target sequence of words associated with a target natural language, based on at least one translation function of attributes associated with candidate translational phrases associated with the communication object.

14. The computer program product of claim 12, wherein the executable code, when executed, is configured to cause the at least one data processing apparatus to:

translate the tokens associated with the first natural language and the tokens included in embedded word phrases associated with the at least one embedded natural language to the target sequence of words associated with the target natural language, based on at least one of;

combining a translation model probability value and a language model probability value that are associated with candidate translational phrases associated with the communication object,

a length of a candidate translation of the group of the object words, and

a difference between a length of a phrase of object words associated with one of the at least one embedded natural language and a length of a phrase associated with the target natural language that corresponds to a candidate translation of the phrase of object words.

15. The computer program product of claim 12, wherein:

the target natural language includes one of the at least one embedded natural language and the first natural language.

16. A method comprising:

receiving a communication object including a plurality of object words;

parsing the communication object to identify each of the object words as tokens;

determining a first natural language and at least one embedded natural language different from the first natural language that are associated with the plurality of object words, based on a language analysis of the tokens; and

translating, via a translating device processor, tokens associated with the first natural language and tokens included in at least one embedded word phrase associated with the at least one embedded natural language to a target natural language, based on at least one context associated with the communication object.

17. The method of claim 16, further comprising:

determining at least one transliterated word included in the tokens, each transliterated word including a transliterated character string; and

converting the at least one transliterated word to a non-transliterated script representation associated with one of the at least one embedded natural language, the first natural language, and the target natural language.

18. The method of claim 16, wherein:

translating the tokens includes translating, via the translating device processor, the tokens associated with the first natural language and the tokens included in embedded word phrases associated with the at least one embedded natural language to a target natural language object, based on at least one translation function of attributes associated with candidate translational phrases associated with the communication object to generate a ranked list of candidate translation phrases.

19. The method of claim 16, wherein:

translating the tokens includes translating, via the translating device processor, the tokens associated with the first natural language and the tokens included in the at least one embedded word phrase associated with the at least one embedded natural language to the target natural language object, based on at least one of;

multiplying a translation model probability value by a language model probability value that are each associated with the communication object,

determining a length of a translation, and

determining a difference between a length of a phrase of object words associated with the at least one embedded natural language and a length of a phrase associated with the target natural language that corresponds to a candidate translation of the phrase of object words.

20. The method of claim 16, wherein:

the target natural language is determined based on a received user input specifying the target natural language.