Query language identification

• US 8,442,965 B2
• Filed: 04/19/2007
• Issued: 05/14/2013
• Est. Priority Date: 04/19/2006
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Lopiano, Fabio
• Primary Examiner(s): Smith, Brannon W
• Application Number:
  US11/737,639
• Publication Number:
  US 20110231423A1
• Time in Patent Office: 2,217 Days
• Field of Search: 707/4, 707/999.004, 707/760, 707/769, 707/706, 707/728, 704/8
• US Class Current: 707/706
• CPC Class Codes: G06F 16/2452 Query translation
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 40/263 Language identification

Abstract

Methods, systems, and apparatus, including computer program products, for identifying the language of a search query. In one embodiment, the language of each term of a query is determined from the query terms and the language of the user interface a user used to enter the query. In another embodiment, an automatic interface language classifier is generated from a collection of past queries each submitted by a user. In some embodiments, a score is determined for each of multiple languages, each score indicating a likelihood that the query language is the corresponding one of the multiple languages.

Claims

1. A computer implemented method, comprising:

providing, in a system comprising one or more computers, a plurality of user interfaces through which search queries are received, wherein each user interface is in a respective interface language, and wherein each interface language is a natural language in which a respective user interface presents information;

maintaining a collection of query records, wherein the collection of query records includes distinct subsets of query records, wherein each distinct subset of query records is associated with a respective user interface of the plurality of user interfaces, wherein each query record associates a past query with one or more result documents, and wherein each result document has an associated natural language;

receiving, in the system, through a first user interface of the plurality of user interfaces, a search query comprising one or more query terms; and

determining, by the system, a query language of the search query from the search query, the interface language of the first user interface, and the distinct subset of query records that is associated with the first user interface, the query language being a natural language;

wherein determining the query language of the search query further comprises;

for each of multiple languages,

calculating a first score for each query term and the respective language, each first score indicating the likelihood that the respective query term is in the respective language, wherein the first score is calculated based on a plurality of documents, each document having an associated natural language,

calculating a second score for the respective language, the second score indicating the likelihood that the search query is in the respective language given the interface language of the first user interface through which the search query was received, where the second score is calculated based on the plurality of query records, and

calculating a third score for the respective language, the third score being a combination of the first score for the respective language and the second score for the respective language; and

determining the query language based on the third scores for the multiple languages.

2. The method of claim 1 where calculating the first score includes:

for each query term;

determining the first score based on a number of times each respective query term appears in the plurality documents in the respective language, the plurality of documents being from a corpus of documents, each document in the corpus of documents being labeled as containing content in one of the multiple languages.

3. The method of claim 1, where determining the query language further comprises:

using an interface language classifier to determine a respective likelihood of the languages that the query is in given the user interface that received the query.

4. The method of claim 3, further comprising:

classifying each past query in the collection of query records based on one or more of the following;

(i) the interface language of the user interface through which the past query was received, (ii) the natural language of the one or more result documents associated with the past query, and (iii) the natural language of the one or more result documents that were selected by the user;

generating an initial distribution of the past queries based on the classifying, wherein the initial distribution indicates what proportion of the past queries from the plurality of query records were in a particular language; and

training the interface language classifier using the initial distribution.

5. The method of claim 1, further comprising:

training an interface language classifier using the collection of query records, wherein each query record of the collection includes a past query received by the system from a respective user, information identifying one or more result documents associated with the past query received by the system, and information identifying which of the one or more result documents associated with the past query were selected by the user for the past query, and wherein the training includes;

classifying each past query in the collection based on one or more of the following;

the interface language of the user interface through which the past query was received, the natural language of the one or more result documents associated with the past query, and the natural language of the one or more result documents that were selected by the user; and

tuning the interface language classifier based on queries that were correctly or incorrectly classified by the classifying until the tuned interface language classifier is sufficiently accurate; and

using the trained interface classifier to generate a respective interface classifier vector for each of the plurality of user interfaces.

6. A computer program product, encoded on a non-transitory machine-readable storage device, operable to cause data processing apparatus to perform operations comprising:

providing, in a system comprising one or more computers, a plurality of user interfaces through which search queries are received, wherein each user interface is in a respective interface language, and wherein each interface language is a natural language in which a respective user interface presents information;

maintaining a collection of query records, wherein the collection of query records includes distinct subsets of query records, wherein each distinct subset of query records is associated with a respective user interface of the plurality of user interfaces, wherein each query record associates a past query with one or more result documents, and wherein each result document has an associated natural language;

receiving, in the system, through a first user interface of the plurality of user interfaces, a search query comprising one or more query terms; and

determining, by the system, a query language of the search query from the search query, the interface language of the first user interface, and the distinct subset of query records that is associated with the first user interface, the query language being a natural language;

wherein determining the query language of the search query further comprises;

for each of multiple languages,

calculating a first score for each query term and the respective language, each first score indicating the likelihood that the respective query term is in the respective language, wherein the first score is calculated based on a plurality of documents, each document having an associated natural language,

calculating a second score for the respective language, the second score indicating the likelihood that the search query is in the respective language given the interface language of the first user interface through which the search query was received, where the second score is calculated based on the plurality of query records, and

calculating a third score for the respective language, the third score being a combination of the first score for the respective language and the second score for the respective language; and

determining the query language based on the third scores for the multiple languages.

7. The product of claim 6 where calculating the first score includes:

for each query term;

determining the first score based on a number of times each respective query term appears in the plurality documents in the respective language, the plurality of documents being from a corpus of documents, each document in the corpus of documents being labeled as containing content in one of the multiple languages.

8. The product of claim 6, where determining the query language further comprises:

using an interface language classifier to determine a respective likelihood of the languages that the query is in given the user interface that received the query.

9. The product of claim 8, further comprising:

classifying each past query in a collection of query records based on one or more of the following;

(i) the interface language of the user interface through which the past query was received, (ii) the natural language of the one or more result documents associated with the past query, and (iii) the natural language of the one or more result documents that were selected by the user;

generating an initial distribution of the past queries based on the classifying, wherein the initial distribution indicates what proportion of the past queries from the plurality of query records were in a particular language; and

training the interface language classifier using the initial distribution.

10. The product of claim 6, further comprising:

training an interface language classifier using the collection of query records, wherein each query record of the collection includes a past query received by the system from a respective user, information identifying one or more result documents associated with the past query received by the system, and information identifying which of the one or more result documents associated with the past query were selected by the user for the past query, and wherein the training includes;

classifying each past query in the collection based on one or more of the following;

the interface language of the user interface through which the past query was received, the natural language of the one or more result documents associated with the past query, and the natural language of the one or more result documents that were selected by the user; and

tuning the interface language classifier based on queries that were correctly or incorrectly classified by the classifying until the tuned interface language classifier is sufficiently accurate; and

using the trained interface classifier to generate a respective interface classifier vector for each of the plurality of user interfaces.

11. A system comprising:

a plurality of user interfaces through which search queries are received, wherein each user interface is in a respective interface language, and wherein each interface language is a natural language in which a respective user interface presents information;

a collection of query records, wherein the collection of query records includes distinct subsets of query records, wherein each distinct subset of query records is associated with a respective user interface of the plurality of user interfaces, wherein each query record associates a past query with one or more result documents, and wherein each result document has an associated natural language;

one or more computers configured to perform operations comprising;

receiving, through the first user interface, a search query comprising one or more query terms;

determining, by the system, a query language of the search query from the search query, the interface language of the first user interface, and the distinct subset of query records that is associated with the first user interface, the query language being a natural language;

wherein determining the query language of the search query further comprises;

for each of multiple languages,

calculating a first score for each query term and the respective language, each first score indicating the likelihood that the respective query term is in the respective language, where the first score is calculated based on a plurality of documents, each document having an associated natural language,

calculating a second score for the respective language, the second score indicating the likelihood that the search query is in the respective language given the interface language of the first user interface through which the search query was received, where the second score is calculated based on the plurality of query records, and

calculating a third score for the respective language, the third score being a combination of the first score for the respective language and the second score for the respective language; and

determining the query language based on the third scores for the multiple languages.

12. The system of claim 11 where calculating the first score includes:

for each query term;

determining the first score based on a number of times each respective query term appears in the plurality documents in the respective language, the plurality of documents being from a corpus of documents, each document in the corpus of documents being labeled as containing content in one of the multiple languages.

13. The system of claim 11, where determining the query language further comprises:

using an interface language classifier to determine a respective likelihood of the languages that the query is in given the user interface that received the query.

14. The system of claim 13, wherein the operations further comprise:

classifying each past query in a collection of query records based on one or more of the following;

(i) the interface language of the user interface through which the past query was received, (ii) the natural language of the one or more result documents associated with the past query, and (iii) the natural language of the one or more result documents that were selected by the user;

generating an initial distribution of the past queries based on the classifying, wherein the initial distribution indicates what proportion of the past queries from the plurality of query records were in a particular language; and

training the interface language classifier using the initial distribution.

15. The system of claim 11, wherein the operations further comprise:

training an interface language classifier using the collection of query records, wherein each query record of the collection includes a past query received by the system from a respective user, information identifying one or more result documents associated with the past query received by the system, and information identifying which of the one or more result documents associated with the past query were selected by the user for the past query, and wherein the training includes;

classifying each past query in the collection based on one or more of the following;

the interface language of the user interface through which the past query was received, the natural language of the one or more result documents associated with the past query, and the natural language of the one or more result documents that were selected by the user; and

tuning the interface language classifier based on queries that were correctly or incorrectly classified by the classifying until the tuned interface language classifier is sufficiently accurate; and

using the trained interface classifier to generate a respective interface classifier vector for each of the plurality of user interfaces.