Correlation across languages

• US 9,342,503 B1
• Filed: 03/12/2013
• Issued: 05/17/2016
• Est. Priority Date: 03/12/2013
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Yang, Dong, Evans, David Kirk
• Primary Examiner(s): ZHU, RICHARD Z
• Application Number:
  US13/797,770
• Time in Patent Office: 1,162 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3332 Query translation
• CPC Class Codes: G06F 16/3337 Translation of the query language, e.g. Chinese to English
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: G06F 40/42 Data-driven translation
• 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 40/53 Processing of non-Latin text kana-to-kanji conversion G06F40/129; vowelisation G06F40/232
• CPC Class Codes: G06F 9/454 Multi-language systems; Localisation; Internationalisation

Abstract

Proper nouns may be correlated across languages by identifying a proper noun term in a first language and identifying a translation associated with the proper noun term, when available. A determination may be made as to whether one or more data stores associates the proper noun term with the translation, and the translation may be correlated with the proper noun term based on the determination.

Claims

1. A method of correlating a proper noun term across languages, comprising:

identifying the proper noun term and a proper noun identifier in a first language from a proper noun term data store to be correlated to a second language using a processor;

determining whether a translation of the proper noun term is available in the proper noun term data store;

determining whether the translation of the proper noun term is available in a product data store;

determining whether the translation of the proper noun term is available in a third-party data store;

translating the proper noun term using transliteration to create the translation when the translation is unavailable;

identifying a translated noun identifier when the translation is available in the proper noun term data store, the product data store or the third-party data store, or when the translation is available as a result of the translating the proper noun term; and

assigning the translation to the proper noun term in the product data store by associating the proper noun identifier with the translated noun identifier in a correlation data store.

2. The method of claim 1, wherein transliterating the proper noun term further comprises transliterating the proper noun term to a phonetic representation of the proper noun term.

3. The method of claim 2, further comprising transliterating the phonetic representation of the proper noun term into the second language.

4. The method of claim 1, wherein the proper noun term is at least one of an author name, an actor name, a book title, a movie title, a song title, a brand name, a product name and an artist name.

5. The method of claim 1, further comprising receiving a search request in one of the first or second languages for the proper noun term.

6. The method of claim 1, further comprising transliterating a search term in a search request from the second language to a phonetic representation of the search term for comparison to at least one of the proper noun term or the transliteration of the proper noun term in the first language.

7. A method of correlating a proper noun term across languages, comprising:

identifying the proper noun term in a first language from a proper noun term data store to be correlated to a second language using a processor;

identifying a translation associated with the proper noun term, the translation being of the proper noun term from the first language into the second language;

comparing search trends for the proper noun term and the translation over time;

determining whether a search trend for the proper noun term approximates a search trend for the translation of the proper noun term to obtain a search trend comparison result; and

associating the translation with the proper noun term based on the search trend comparison result.

8. The method of claim 7, wherein identifying the translation associated with the proper noun term further comprises identifying whether the translation is available in a proper noun term data store.

9. The method of claim 7, wherein identifying the translation associated with the proper noun term further comprises identifying whether the translation is available in a second data store.

10. The method of claim 9, wherein the second data store is a remote, public data store and the first data store is a local, private data store.

11. The method of claim 7, wherein associating the translation with the proper noun term comprises associating the translation with the proper noun term from a second data store.

12. The method of claim 11, wherein the first data store is a remote, public data store and the second data store is a local, private data store.

13. The method of claim 8, wherein the proper noun term data store includes an identifier for each proper noun term, the method further comprising searching for the proper noun term in a first data store by searching for the identifier.

14. The method of claim 7, further comprising transliterating the proper noun term when the translation is not available.

15. The method of claim 14, wherein transliterating further comprises using a statistical transliteration approach for transliterating proper noun terms.

16. The method of claim 7, further comprising transliterating the proper noun term to a phonetic representation of the proper noun term.

17. The method of claim 16, further comprising transliterating the phonetic representation of the proper noun term into the second language.

18. The method of claim 7, further comprising transliterating the proper noun term to obtain a candidate transliteration and ranking the candidate transliteration of the proper noun term with other candidate transliterations for use in obtaining search results by comparing the search trends of the candidate transliteration and the other candidate transliterations with the search trend for the proper noun term.

19. The method of claim 7, wherein the translation is one of a plurality of transliterations, the method further comprises tracking search frequencies for the plurality of transliterations to identify the search trends and selecting the one of the plurality of transliterations as the translation based on a search frequency for the one of the plurality of transliterations as compared with the plurality of transliterations.

20. The method of claim 7, wherein the proper noun term is at least one of an author name, an actor name, a book title, a movie title, a song title, a brand name, a product name and an artist name.

21. The method of claim 7, further comprising receiving a search request in the second language for the proper noun term.

22. The method of claim 21, further comprising transliterating a search term in the second language to a phonetic representation of the search term for comparison to at least one proper noun term and the translation.

23. A method of correlating proper noun terms across languages, comprising:

identifying an identifier associated with a proper noun term to be correlated across a plurality of languages from a proper noun term data store using a processor;

searching for the identifier in a first data store associated with a first language and in a second data store associated with a second language using a correlation data store as an index;

identifying the proper noun term in the first language from the first data store and the proper noun term in the second language from the second data store based on correlation of the identifier in the first and second data stores from the correlation data store;

translating the proper noun term into the second language when the translation is unavailable in the second data store; and

associating the translation with the identifier in the correlation data store.

24. The method of claim 23, further comprising receiving a search request in the second language for the proper noun term and providing search results in the first language.

25. A method of correlating proper noun terms across languages, comprising:

identifying a proper noun term in a first language from a search query received from a user to be correlated with the proper noun term in a second language using a processor;

determining whether the proper noun term in the first language is associated with the proper noun term in the second language;

comparing search trends for the proper noun term and a translation candidate of the proper noun term in the first language over time to determine whether a search trend for the proper noun term approximates a search trend for the translation candidate to obtain a search trend comparison result; and

providing search results for the translation candidate based on the search trend comparison result.