Translating natural language utterances to keyword search queries

• US 9,064,006 B2
• Filed: 08/23/2012
• Issued: 06/23/2015
• Est. Priority Date: 08/23/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Hakkani-Tur, Dilek Zeynep, Tur, Gokhan, Iyer, Rukmini, Heck, Larry Paul
• Primary Examiner(s): LEROUX, ETIENNE PIERRE
• Application Number:
  US13/592,638
• Publication Number:
  US 20140059030A1
• Time in Patent Office: 1,034 Days
• Field of Search: 707/5, 707/706, 709204-206
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24522 Translation of natural language queries to structured queries
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3332 Query translation
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/84 Mapping; Conversion
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 40/44 Statistical methods, e.g. probability models
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

Natural language query translation may be provided. A statistical model may be trained to detect domains according to a plurality of query click log data. Upon receiving a natural language query, the statistical model may be used to translate the natural language query into an action. The action may then be performed and at least one result associated with performing the action may be provided.

Claims

1. A method for providing natural language query translation, the method comprising:

training a statistical model according to a plurality of query click log data, the plurality of click log data being mined to train the statistical model for domain detection in the absence of available in-domain data;

receiving a natural language query;

translating the natural language query into a search query according to the statistical model;

performing the search query; and

providing at least one result associated with performing the search query.

2. The method of claim 1, wherein the natural language query is received as text.

3. The method of claim 1, wherein the natural language query is received as speech.

4. The method of claim 1, wherein training the statistical model comprises identifying a plurality of domain independent salient phrases.

5. The method of claim 4, wherein each of the plurality of domain independent salient phrases comprises at least one word indicating that an associated search query comprises a natural language search query.

6. The method of claim 4, wherein training the plurality of query click log data is associated with a plurality of search engine results.

7. The method of claim 6, further comprising identifying a plurality of search queries associated with the plurality of query click log data that comprise natural language search queries according to the plurality of domain independent salient phrases.

8. The method of claim 7, further comprising generating a query pair by correlating at least one of the plurality of natural language search queries to at least one keyword-based search query.

9. The method of claim 8, wherein the correlation between the at least one of the plurality of natural language search queries and the at least one keyword-based search query is associated with a Uniform Resource Locator (URL) distribution.

10. The method of claim 9, wherein performing the search query comprises:

searching the plurality of query click log data for a query pair corresponding to the search query; and

identifying a domain associated with the search query according to the URL distribution.

11. A system for providing natural language query translation, the system comprising:

a memory storage; and

a processing unit coupled to the memory storage, wherein the processing unit is operable to;

receive a query from a user,

determine whether the query comprises a natural language query, and

in response to determining that the query comprises the natural language query;

map the natural language query into a keyword-based query, wherein being operative to map the natural language query into the key-board based query comprises being operative to detect a domain associated with the natural language query utilizing mined click data in the absence of available in-domain data;

perform a search according to a query pair comprising the natural language query and the keyword-based query; and

provide a plurality of results associated with the search to the user.

12. The system of claim 11, wherein being operative to map the natural language query into the keyword-based query comprises being operative to:

detect a domain associated with the natural language query; and

strip at least one domain-independent word from the natural language query.

13. The system of claim 12, wherein being operative to detect the domain associated with the natural language query comprises being operative to:

identify a subset of a plurality of possible domains according to at least one feature of the natural language query.

14. The system of claim 13, wherein the at least one feature of the natural language query comprises at least one of the following:

a lexical feature, a contextual feature, a semantic feature, a syntactic feature, and a topical feature.

15. The system of claim 13, wherein being operative to map the natural language query into the keyword-based query comprises being further operative to:

convert the natural language query into the keyword-based query according to a trained statistical machine translation model.

16. The system of claim 15, wherein the statistical machine translation model is trained according to a plurality of mined query pairs each comprising a previous natural language query and an associated previous keyword-based query.

17. The system of claim 16, wherein the previous natural language query is identified according to a domain independent salient phrase.

18. The system of claim 16, wherein the previous natural language query and the previous keyword-based query are associated according to a weighted Uniform Resource Locator (URL) click graph.

19. A computer-readable medium which stores a set of instructions which when executed performs a method for providing natural language query translation, the method executed by the set of instructions comprising:

training a statistical machine translation model according to a plurality of mined query pairs, wherein training the statistical machine translation model comprises;

identifying a plurality of domain independent salient phrases (DISPs),

identifying a plurality of previous natural language queries according to the plurality of DISPs,

associating each of the plurality of previous natural language queries with a previous keyword-based query into a mined query pair of the plurality of mined query pairs according to a uniform resource locator (URL) click graph, wherein the URL click graph comprises a weighted distribution of URLs selected in response to the a previous natural language queries and previous keyword-based queries, the URL click graph comprising a bi-partite query quick graph having nodes corresponding to the natural language queries and the URLs; and

extracting a plurality of common features for each of the mined query pairs;

receiving a new query from a user,

determining whether the new query comprises a new natural language query,

in response to determining that the query comprises the natural language query, mapping the new natural language query into a keyword-based query according to the trained statistical machine translation model;

performing a search according to the new query; and

providing a plurality of results associated with the search to the user.