Automatic extraction of multilingual dictionary items from non-parallel, multilingual, semi-structured data

• US 9,805,031 B2
• Filed: 12/13/2016
• Issued: 10/31/2017
• Est. Priority Date: 02/28/2014
• Status: Expired due to Fees

• Current Assignee: eBay Inc.
• Original Assignee: eBay Inc.
• Inventors: Sawaf, Hassan
• Primary Examiner(s): ZHU, RICHARD Z
• Application Number:
  US15/377,602
• Publication Number:
  US 20170091176A1
• Time in Patent Office: 322 Days
• Field of Search: None
• CPC Class Codes: G06F 16/903 Querying for retrieval from the web G06F16/953
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G06F 40/42 Data-driven translation
• CPC Class Codes: G06F 40/44 Statistical methods, e.g. probability models
• CPC Class Codes: G06F 40/45 Example-based machine translation; Alignment
• CPC Class Codes: G06F 40/49 using very large corpora, e.g. the web
• CPC Class Codes: G06F 40/51 Translation evaluation
• 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
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06Q 30/0601 Electronic shopping [e-shopping]
• CPC Class Codes: G06Q 30/0609 Buyer or seller confidence or verification

Abstract

User interfaces receive a first plurality of user queries and result sets that are in a category, exhibit a constraint, and exhibit user behavior. Also received are a second plurality of user queries and result sets that that are in the category, exhibit the constraint, and exhibit user behavior. The second user queries and results are received either from a plurality of user interfaces coupled to a second system, or from the second system itself. Responsive to detecting that the first plurality of user queries and result sets and the second plurality of user queries and result sets satisfy respective thresholds, a signal indicates that at least one of the first plurality of user queries and at least one of the second plurality of user queries are translations of each other.

Claims

1. A method of machine translation comprising:

receiving at a second publication system, from a first publication system, in a first language, a first plurality of user queries for a product, and a first plurality of result sets in the first language that are responsive to the first plurality of user queries, respective ones of the first plurality of result sets evoking a respective set of implicit user behaviors;

receiving, at the first publication system, in a second language, from the second publication system, a second plurality of user queries for a product, and a second plurality of result sets in the second language that are responsive to the second plurality of user queries, respective ones of the second plurality of result sets evoking a respective set of implicit user behaviors;

detecting a number of the first plurality of result sets that evokes implicit positive user behaviors;

detecting a number of the second plurality of result sets that evokes implicit positive user behaviors;

detecting that the number of the first plurality of result sets that evokes implicit positive user behaviors satisfies a first threshold and the number of the second plurality of result sets that evokes implicit positive user behaviors satisfies a second threshold; and

responsive to detecting that the number of first plurality of result sets that evokes implicit positive user behaviors satisfies the first threshold and the number of second plurality of result sets that evokes implicit positive user behaviors satisfies the second threshold, providing an indication that a query of the first plurality of user queries and a query of the second plurality of user queries are likely to be translations of each other.

2. The method of claim 1 wherein the indication is provided to a machine translation application for the second publication system and to a machine translation application for the first publication system.

3. The method machine language translation of claim 1 wherein the first plurality of user queries in the first language are in a category and evoke a first constraint, and the second plurality of user queries in the second language are in the category and evoke a second constraint that is similar to the first constraint.

4. The method of machine translation of claim 3 wherein the constraint comprises one of an aspect, a bar code or a seller.

5. The method of machine translation of claim 3 wherein the constraint comprises similar product features.

6. The method of machine translation of claim 3 wherein the categories are the same category.

7. The method of machine translation of claim 3 wherein the constraint comprises a product image.

8. One or more computer-readable hardware storage device having stored therein a set of instructions which, when executed by one or more processors of a computer, causes the computer to execute operations comprising:

receiving at a second publication system, in a first language, from a first publication system, a first plurality of user queries for a product, and a first plurality of result sets in the first language that are responsive to the first plurality of user queries, respective ones of the first plurality of result sets evoking a respective set of implicit user behaviors;

receiving, at the first publication system, in a second language, from the second publication system, a second plurality of user queries for a product, and a second plurality of result sets in the second language that are responsive to the second plurality of user queries, respective ones of the second plurality of result sets evoking a respective set of implicit user behaviors;

detecting a number of the first plurality of result sets that evokes implicit positive user behaviors;

detecting a number of the second plurality of result sets that evokes implicit positive user behaviors;

detecting that the number of the first plurality of result sets that evokes implicit positive user behaviors in response to the first plurality of result sets satisfies a first threshold and the number of the second plurality of result sets that evokes implicit positive user behaviors in response to the second plurality of result sets satisfies a second threshold; and

responsive to detecting that the number of the first plurality of result sets that evokes implicit positive user behaviors satisfies the first threshold and the number of the second plurality of result sets that evokes implicit positive user behaviors satisfies the second threshold, providing an indication that a query of the first plurality of user queries and a query of the second plurality of user queries are likely to be translations of each other.

9. The method of claim 8 wherein the indication is provided to a machine translation application for the second publication system and to a machine translation application for the first publication system.

10. The one or more computer-readable hardware storage device of claim 8 wherein the first plurality of user queries in the first language are in a category and evoke a first constraint, and the second plurality of user queries in the second language are in the category and evoke a second constraint that is similar to the first constraint.

11. The method of machine translation of claim 10 wherein the constraint comprises one of an aspect, a bar code or a seller.

12. The one or more computer-readable hardware storage device of claim 10 wherein the constraint comprises similar product features.

13. The one or more computer-readable hardware storage device of claim 10 wherein the categories are the same category.

14. The one or more computer-readable hardware storage device of claim 10 wherein the constraint comprises a product image.

15. A system comprising one or more hardware processors configured to:

receive at a second publication system, from a first publication system, in a first language, a first plurality of user queries for a product, and a first plurality of result sets in the first language, that are responsive to the first plurality of user queries, wherein respective ones of the first plurality of result sets evoke a respective set of implicit user behaviors;

receive, at the first publication system, from the second publication system, in a second language, a second plurality of user queries for a product in a second language, and a second plurality of result sets in the second language, that are responsive to the second plurality of user queries, wherein respective ones of the second plurality of result sets evoke a respective set of implicit user behaviors;

detect, by a user behavior module of a machine translation application at the second publication system, a number of the first plurality of result sets that evokes implicit positive user behaviors;

detect, by a user behavior module of a machine translation application at the first publication system, a number of the second plurality of result sets that evokes implicit positive user behaviors;

detect, by a decision application, that the number of the first plurality of result sets that evokes implicit positive user behaviors satisfies a first threshold and the number of the second plurality of result sets that evokes implicit positive user behaviors satisfies a second threshold; and

responsive to detection that the number of the first plurality of result sets that evokes implicit positive user behaviors satisfies the first threshold and the number of the second plurality of result sets that evokes implicit positive user behaviors satisfies the second threshold, provide an indication that a query of the first plurality of user queries and a query of the second plurality of user queries are likely to be translations of each other.

16. The system of claim 15 wherein the indication is provided to a machine translation application for the second publication system and to a machine translation application for the first publication.

17. The system of claim 15 wherein the first plurality of user queries in the first language are in a category and evoke a first constraint, and the second plurality of user queries in the second language are in the category and evoke a second constraint that is similar to the first constraint.

18. The system of claim 17 wherein the constraint comprises one of an aspect, a bar code or a seller.

19. The system of claim 17 wherein the constraint comprises similar product features.

20. The system of claim 17 wherein the constraint comprises a product image.