Machine translation split between front end and back end processors

• US 8,209,162 B2
• Filed: 05/01/2006
• Issued: 06/26/2012
• Est. Priority Date: 05/01/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Menezes, Arul A., Quirk, Christopher B., Rashid, Richard F., Moore, Robert C.
• Primary Examiner(s): He, Jialong
• Application Number:
  US11/414,844
• Publication Number:
  US 20070255550A1
• Time in Patent Office: 2,248 Days
• Field of Search: 704 2- 8, 704/10
• 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

A method of translation includes uploading a source text portion to a back end processor. The back end processor identifies a subset of translation knowledge associated with the source text portion. The back end processor downloads the subset to a front end processor. A translation engine runs on the front end processor. The translation engine generates a translation of the source text portion as a function of the subset.

Claims

1. A method of translation, comprising:

uploading a source text portion to be translated to a back end processor that identifies a subset of translation knowledge specifically associated with the uploaded source text portion and downloading the subset; and

running a translation engine on a processor other than the back end processor to generate a translation of the source text portion as a function of the subset,

wherein uploading the source text portion to the back end processor, which identifies the subset of translation knowledge specifically associated with the source text portion, is carried out prior to any attempt to translate the source text portion on the processor other than the back end processor, and

wherein the back end processor splits the source text portion into minimal independent translation segments, and wherein the back end processor identifies, for each minimal independent translation segment of the minimal independent translation segments, a corresponding knowledge segment, and wherein the back end processor assembles the identified knowledge segments corresponding to respective ones of the minimal independent translation segments to form the subset of translation knowledge.

2. The method of claim 1 further comprising:

processing a first number of instructions in the translation engine for the source text portion;

processing a second number of instructions in the back end processor for the source text portion,

wherein the first number is greater than the second number, providing an uneven split of back end processor instruction processing times and processing times on a processor other than the back end processor, the uneven split leveraging the back end processor for multiple tasks.

3. The method of claim 1 wherein the identifying comprises limiting the subset to a limited number of words and phrases potentially in the source text portion.

4. The method of claim 1 wherein the back end processor comprises a web service provider.

5. The method of claim 1 wherein the back end processor comprises a computing cluster.

6. The method of claim 1 wherein the translation engine does not access the back end processor during translation of a source text portion.

7. The method of claim 1 further comprising:

dividing a source text into multiple source text portions and providing the multiple source text portions to the back end processor.

8. The method of claim 1 further comprising:

converting an original file that includes non-translatable features to a source file from which the non-translatable features are removed.

9. The method of claim 8 further comprising:

combining a translation with the non-translatable features to provide a translation that includes the non-translatable features.

10. A computer storage medium having computer executable components, comprising:

a fetch component that fetches a source text portion to be translated to a back end processor, which identifies a subset of translation knowledge specifically associated with the fetched source text portion;

a download component that downloads the subset of translation knowledge associated with the source text portion; and

a translation engine on a processor other than the back end processor that generates a translation of the source text portion depending wholly or in part on the translation knowledge subset,

wherein the fetch component fetches the source text portion to the back end processor, which identifies the subset of translation knowledge specifically associated with the source text portion, prior to any attempt to translate the source text portion on the processor other than the back end processor, and

wherein the back end processor splits the source text portion into minimal independent translation segments, and wherein the back end processor identifies, for each minimal independent translation segment of the minimal independent translation segments, a corresponding knowledge segment, and wherein the back end processor assembles the identified knowledge segments corresponding to respective ones of the minimal independent translation segments to form the subset of translation knowledge.

11. The medium of claim 10 wherein:

the translation engine processes a first number of instructions for the source text portion;

the back end processor processes a second number of instructions for the source text portion, and

wherein the first number is greater than the second number, providing an uneven split of processor instruction processing times that leverages the back end processor for multiple tasks.

12. The medium of claim 10 wherein the translation engine does not access the back end processor during translation of a source text portion.

13. The medium of claim 10 further comprising:

a component that divides a source text into multiple source text portions in the front end processor and uploads the multiple source text portions to the back end processor.

14. The medium of claim 10 further comprising:

a component that converts an original file that includes non-translatable features to a source file from which the non-translatable features are removed.

15. The medium of claim 14 further comprising:

a component that combines the translation with the non-translatable features to provide a translation that includes the non-translatable features.

16. A method comprising:

dividing, based on a specified pre-determined length, a source text to be translated into multiple source text portions;

uploading each respective one of the multiple source text portions to a back end processor before any attempt is made to translate the corresponding respective one of the multiple source text portions;

identifying, using the back end processor, a corresponding translation knowledge subset for each of the multiple source text portions;

downloading the corresponding translation knowledge subset for each of the multiple source text portions; and

running a translation engine on a processor other than the back end processor to generate a translation for each of the multiple source text portions as a function of the corresponding translation knowledge subset for each of the multiple source text portions,

wherein the back end processor splits each source text portion of the multiple source text portions into minimal independent translation segments, and wherein the back end processor identifies, for each minimal independent translation segment of the minimal independent translation segments, a corresponding knowledge segment, and wherein the back end processor assembles the identified knowledge segments corresponding to respective ones of the minimal independent translation segments to form the corresponding translation knowledge subset.

17. The method of claim 16 wherein the back end processor comprise a web service provider.

18. The method of claim 16 wherein the back end processor comprises a computing cluster.

19. The method of claim 16 wherein the translation engine does not access the back end processor during translation of the multiple source text portions.