Automated evaluation and selection of machine translation protocols

Patent metadata

Field	Value
Publication number	US-11449686-B1
Application number	US-201916506354-A
Country	US
Kind code	B1
Filing date	Jul 9, 2019
Priority date	Jul 9, 2019
Publication date	Sep 20, 2022
Grant date	Sep 20, 2022

Abstract

Systems, devices, and methods are provided for using an automated assessment and evaluation of machine translations. A system may receive a request associated with translating first content from a first language to a second language. The system may translate the first content from the first language to the second language. The system may determine an attribute associated with the first content. The system may determine a translation score associated with second content translated from the first language to the second language and associated with the attribute, the translation score indicative of a machine translation accuracy. The system may determine, and based on the translation score, a translation protocol, and may execute the translation protocol.

First claim

That which is claimed is:

1. A method, comprising:
receiving, by at least one processor coupled to memory, a request associated with translating first content from a first language to a second language;
determining, by the at least one processor, a translation of the first content from the first language to the second language;
determining, by the at least one processor, an attribute associated with the translation of the first content from the first language to the second language;
determining, by the at least one processor, based on the attribute, a first estimated time to translate the first content from the first language to the second language using a first machine translation protocol;
determining, by the at least one processor, based on the attribute, a second estimated time to translate the first content from the first language to the second language using a second machine translation protocol;
determining that the first estimated time is less than the second estimated time;
selecting, by the at least one processor and based on the first estimated time being less than the second estimated time, the first machine translation protocol; and
executing the first machine translation protocol.

2. The method of claim 1 , wherein executing the first machine translation protocol comprises:
sending a request associated with a human operator review of the translation; and
receiving a modified version of the translation.

3. The method of claim 1 , wherein executing the first machine translation protocol comprises sending the translation, and wherein the first machine translation protocol comprises a machine-only translation protocol.

4. The method of claim 1 , further comprising:
determining, based on the attribute, a score threshold;
generating a translation score associated with second content previously translated from the first language to the second language and associated with the attribute, and wherein the translation score is indicative of a machine translation accuracy; and
comparing the translation score to the score threshold, wherein selecting the first machine translation protocol is further based on the comparison.

5. The method of claim 1 , further comprising:
determining, based on the attribute, a first estimated accuracy associated with a first human operator review;
determining, based on the attribute, a second estimated accuracy associated with a second human operator review; and
determining that the first estimated accuracy is greater than the second estimated accuracy, wherein the first machine translation protocol comprises the first human operator review.

6. The method of claim 1 , further comprising:
determining, based on the attribute, a first estimated cost to translate the first content from the first language to the second language using a first machine translation protocol;
determining, based on the attribute, a second estimated cost to translate the first content from the first language to the second language using a second machine translation protocol; and
determining that the first estimated cost is less than the second estimated cost, wherein selecting the first machine translation protocol is further based on the first estimated cost being less than the second estimated cost.

7. The method of claim 1 , further comprising:
determining, based on the attribute, a product associated with the first content;
determining an identifier associated with the product;
determining, based on the identifier, that second content previously translated from the first language to the second language is associated with the product; and
determining that the second content is associated with the attribute.

8. The method of claim 1 , further comprising:
determining a type of content associated with the first content; and
determining that second content previously translated from the first language to the second language is associated with the type of content.

9. The method of claim 1 , wherein the attribute is based on a number of words in the first content.

10. The method of claim 1 , further comprising receiving a product review, wherein the attribute is associated with the product review.

11. The method of claim 1 , further comprising generating a first translation score associated with second content previously translated from the first language to the second language and associated with the attribute, wherein the first translation score is indicative of a machine translation accuracy, wherein the first translation score is based on a human operator review of the second content in the second language, further comprising determining a second translation score associated with the second content, wherein the second translation score is based on a computer-based evaluation of the second content in the second language, and wherein selecting the first machine translation protocol is further based on the second translation score.

12. The method of claim 1 , further comprising
generating a first translation score associated with second content previously translated from the first language to the second language and associated with the attribute, and wherein the first translation score is indicative of a machine translation accuracy;
determining a second translation score associated with the translation of the first content from the first language to the second language; and
determining, based on the second translation score, an adjustment associated with the first translation score.

13. The method of claim 1 , wherein the request is a first request, wherein the attribute is a first attribute, and wherein the translation is a first translation, further comprising:
receiving a second request associated with translating third content from the first language to the second language;
determining a second translation of the third content from the first language to the second language using the second machine translation protocol;
determining a second attribute associated with the second translation;
determining an absence of content associated with the second attribute; and
executing the second machine translation protocol, wherein the second machine translation protocol comprises a human operator evaluation of the second translation.

14. A system comprising memory and at least one processor, the at least one processor configured to:
receive a request associated with translating first content from a first language to a second language;
determine a translation of the first content from the first language to the second language;
determine an attribute associated with the translation;
determine, based on the attribute, a first estimated cost to translate the first content from the first language to the second language using a first machine translation protocol;
determine, based on the attribute, a second estimated cost to translate the first content from the first language to the second language using a second machine translation protocol; and
determine that the first estimated cost is less than the second estimated cost;
select, based on the first estimated cost being less than the second estimated cost, the first machine translation protocol; and
execute the first machine translation protocol.

15. The system of claim 14 , wherein to execute the first machine translation protocol comprises the at least one processor being further configured to:
send a request associated with a human operator review of the translation; and
receive a modified version of the translation.

16. A method, comprising:
receiving, by at least one processor coupled

Assignees

• Amazon Tech Inc

Inventors

• Panchal Brinda
• Agarwal Abhinav
• Barnabas Sandy
• Ling You
• Fomitchov Pavel
• Vasudevamurthy Nagaraja
• Wadia Kaivan
• Lamptey Emmanuel Addy
• Kang Meng

Classifications

• G06F40/58 Primary

  Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation · CPC title

• G06F40/51 Primary

  Translation evaluation · CPC title