Customizable machine translation service

• US 8,831,928 B2
• Filed: 04/04/2007
• Issued: 09/09/2014
• Est. Priority Date: 04/04/2007
• Status: Active Grant

• Current Assignee: SDL Inc (RWS Holdings Plc)
• Original Assignee: Language Weaver, Inc. (RWS Holdings Plc)
• Inventors: Marcu, Daniel, Wong, William, Lung, Felix
• Primary Examiner(s): Godbold, Douglas
• Assistant Examiner(s): Ortiz Sanchez, Michael
• Application Number:
  US11/784,161
• Publication Number:
  US 20080249760A1
• Time in Patent Office: 2,715 Days
• Field of Search: 704 1- 10
• US Class Current: 704/2
• 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

Embodiments of the present invention provide a system and method for providing a translation service. The method comprises providing a translation interface accessible via a network. The translation interface receives specialized data associated with a domain from a member. A text string written in a source language is received from the member via the translation interface. A domain-based translation engine is selected. The domain-based translation engine may be associated with a source language, a target language, and a domain. The text string is translated into the target language using, at least in part, the selected domain-based translation engine. The translated text string is transmitted to the member via the Internet. In some embodiments, a translation memory is generated based on the specialized data.

Claims

1. A method for providing a translation service comprising:

receiving a text string written in a source language from a member via a translation interface;

receiving specialized data associated with;

a domain from the member via the translation interface, and

an indication of confidentiality;

generating a translation memory using the received specialized data;

generating a set of domain-specific parameters for statistical machine translation using the received specialized data;

merging the generated set of domain-specific parameters with a generic set of statistical machine translation parameters;

generating a domain-based translation engine based on the received specialized data and generic data, the domain-based translation engine associated with a source language, a target language, and the domain;

training the domain-based translation engine using the translation memory;

receiving a selection of the generated domain-based translation engine from a plurality of domain-based translation engines, the selection received from the member via the translation interface;

translating the text string into the target language using, at least in part, the selected domain-based translation engine, the merged parameters and translation memory being used by the domain-based translation engine for translating the text string; and

transmitting the translated text string to the member via the translation interface over a network.

2. The method of claim 1, wherein the translation interface is accessible via the network, the translation interface configured to receive translation memory associated with the domain from another member.

3. The method of claim 1, further comprising sending the generated translation memory to the member for storage.

4. The method of claim 1, further comprising:

translating a first portion of the text string using the translation memory;

annotating the translation of the first portion of the text string with a first confidence level;

translating a second portion of the text string using the merged parameters; and

annotating the translation of the second portion of the text string with a second confidence level.

5. The method of claim 1, further comprising:

determining if the specialized data includes a sufficient amount of data for generating a domain-based translation engine; and

generating the domain-based translation engine based on the determination that the specialized data includes a sufficient amount of data and a requirement received from the member to generate the domain-based translation engine using the specialized data.

6. The method of claim 1, further comprising:

translating a second text string received from a second member using, at least in part, specialized data received from the second member; and

crediting an account associated with the second member.

7. The method of claim 1, further comprising receiving a payment from the member based on a subscription.

8. The method of claim 1, further comprising receiving a payment from the member based on a usage fee.

9. The method of claim 1, further comprising providing to the member confidence level information based on the merged parameters and associated with the translated text string.

10. The method of claim 1, wherein translating the text string is performed using, at least in part, a data-driven machine translation technique.

11. A domain-based translation engine comprising:

a communications module configured to receive specialized data and to receive a source language text string written in a source language and to receive a target language text string written in a target language, from a member via a network;

a translation module configured to translate the source language text string into the target language and to translate the target language text string into the source language using, at least in part, data-driven machine translation and the specialized data, wherein the domain-based translation engine is generated based on received domain data; and

a customization engine configured to determine whether an amount of the specialized data is sufficient to generate the domain-based translation engine and generate a language memory using the specialized data if the amount of the specialized data is not sufficient or to generate the domain-based translation engine if the amount of the specialized data is sufficient.

12. The domain-based translation engine of claim 11, wherein the translation module is communicatively coupled with a translation memory engine configured to generate translation memory used for training the domain-based translation engine and accessed by the domain-based translation engine to translate the input text string.

13. A computer readable non-transitory medium having embodied thereon a program, the program being executable by a processor for performing a method for providing a translation service, the method comprising:

receiving a text string written in a source language from a member via a translation interface;

receiving translation memory from the member, the translation memory associated with the source language, a target language, and a domain;

receiving specialized data associated with the domain, the source language, and the target language, the specialized data including a set of domain-based statistical machine translation parameters;

merging the domain-based statistical machine translation parameters and a set of generic statistical machine translation parameters;

determining if the specialized data includes a sufficient amount of data for generating a domain-based translation engine;

receiving instructions from the member to generate the domain-based translation engine using the translation memory;

generating the domain-based translation engine based on the determination that the specialized data includes a sufficient amount of data and a requirement received from the member to generate the domain-based translation engine using the specialized data;

training the generated domain-based translation engine using both the translation memory and the specialized data;

receiving a selection of the domain-based translation engine from the member;

translating the text string into the target language using, at least in part, the selected domain-based translation engine, the received translation memory and the merged translation parameters accessed by the domain-based translation engine for translating the text string; and

transmitting the translated text string to the member via the translation interface over an Internet.

14. The computer readable medium of claim 13, wherein the method further comprises:

receiving specialized data associated with a domain from another member; and

training the domain-based translation engine using the specialized data associated with the domain from another member and generic data.

15. The computer readable medium of claim 13, wherein the method further comprises translating at least a portion of the text string using the received translation memory and another portion of the text string using the merged parameters.

16. The computer readable medium of claim 13, wherein the method further comprises generating the domain-based translation engine using a customization engine based on the translation memory received from the member.

17. The computer readable medium of claim 13, wherein the method further comprises:

translating a second text string received from a second member from the target language to the source language using, at least in part, the domain-based translation engine; and

crediting an account associated with the member.

18. The computer readable medium of claim 13, wherein the method further comprises providing confidence level information associated with the parameters used for translating the text string.

19. The computer readable medium of claim 13, wherein specialized data is associated with an indication of confidentiality.

20. A method comprising:

receiving a plurality of documents from a member, a first document of the plurality of documents written in a first source language and a second document of the plurality of documents written in a second source language;

receiving instructions from the member including instructions to use a first domain-based translation engine for translating the first document and instructions to use a second domain-based translation engine for translating the second document;

receiving a set of domain-based statistical machine translation parameters;

merging the domain-based statistical machine translation parameters and a set of generic statistical machine translation parameters;

generating a first translation memory based on the first domain-based translation engine;

sending the first translation memory to the member for storage;

translating a first portion of the first document using the first translation memory based on the first domain-based translation engine and generic data;

annotating the translation of the first portion of the first document with a first confidence level;

translating a second portion of the first document using the merged parameters;

annotating the translation of the second portion of the first document with a second confidence level; and

generating a translation of the second document based on the second domain-based translation engine and generic data.

21. The method of claim 20, further comprising generating the second domain-based translation engine based on a second translation memory and the plurality of documents.

22. The method of claim 21, further comprising translating a text string received from another member using the second domain-based translation engine.

23. The method of claim 20, wherein the first document is associated with a first format and the second document is associated with a second format.