Systems and methods for creating and using geo-centric language models

• US 8,255,217 B2
• Filed: 10/16/2009
• Issued: 08/28/2012
• Est. Priority Date: 10/16/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Stent, Amanda, Zeljkovic, Ilija, Wilpon, Jay, Caseiro, Diamantino
• Primary Examiner(s): Godbold, Douglas
• Application Number:
  US12/580,853
• Publication Number:
  US 20110093265A1
• Time in Patent Office: 1,047 Days
• Field of Search: 704231-245, 704270-2701
• US Class Current: 704/243
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Systems and methods for creating and using geo-centric language models are provided herein. An exemplary method includes assigning each of a plurality of listings to a local service area, determining a geographic center for the local service area, computing a listing density for the local service area, and selecting a desired number of listings for a geo-centric listing set. The geo-centric listing set includes a subset of the plurality of listings. The exemplary method further includes dividing the local service area into regions based upon the listing density and the number of listings in the geo-centric listing set, and building a language model for the geo-centric listing set.

Claims

1. A method, for creating a geo-centric language model for use in performing a search for a user of a mobile communications device, comprising:

determining a location of the mobile communications device;

determining, at a language model creation engine, a geographic center for a local service area, to be used in the search, based on the location of the mobile communications device, yielding a determined local service area;

computing, at the language model creation engine, a listing density, for the local service area, the listing density representing a density of searchable listings located in the determined local service area, yielding a computed listing density,

determining, at the language model creation engine, based on the computed listing density, a radius for use in creating the geo-centric language model, yielding a determined radius; and

determining, at the language model creation engine, the geo-centric language model to correspond to a region defined by the determined radius about the location of the mobile communications device.

2. The method of claim 1, wherein the steps are performed by a voice-driven search system, which includes the language model creation engine.

3. The method of claim 2, wherein the voice-driven search system is one of a network-based voice-driven search system and a device-based voice-driven search system.

4. The method of claim 2 wherein:

the method further comprises receiving, at the voice-driven search system, said location;

determining the qeo-centric language model, corresponding to the region defined by the determined radius about the location of the mobile communications device, includes selecting, at the voice-driven search system, one of a plurality of language models stored in a language model database that covers said region;

the method further comprises receiving, at the voice-driven search system, an audio input provided by the user; and

the method further comprises processing, at the voice-driven search system, the audio input based at least in part upon the determined language model to determine a search request identified in the audio input.

5. The method of claim 4, wherein receiving, at the voice-driven search system, the location comprises receiving the location as determined by the mobile communication device, the mobile device being in communication with the voice-driven search system, wherein the mobile device determines the location via at least one of cellular triangulation, WIFI triangulation, global positioning system (GPS), assisted GPS (A-GPS), area code, and user input.

6. The method of claim 4, wherein receiving, at the voice-driven search system, the location comprises receiving the location as determined by the mobile communication device, the mobile device comprising the voice-driven search system, wherein the mobile device determines the location via at least one of cellular triangulation, WIFI triangulation, GPS, assisted GPS (A-GPS), area code, and user input.

7. The method of claim further comprising the language model creation engine excluding, from consideration in determining the geo-centric language model, any listings that are listed in the region but have an address or geo-coordinates outside of the region.

8. A method for creating geo-centric language models, the method comprising:

computing, at a language model creation engine, a mean and a standard deviation for latitude and longitude coordinates over all listings in a local service area;

computing, at the language model creation engine, a listing density for a most listing-dense region in the local service area;

computing, at the language model creation engine, a geo-location quantization accuracy for the local service area;

quantizing, at the language model creation engine, a set of geo-coordinates for each listing in the local service area, each unique set of geo-coordinates defining a center of a language model;

determining, at the language model creation engine, a number of listings within at least two predetermined radii of the language model center for one of the unique sets of geo-coordinates;

selecting, at the language model creation engine, a radius from the at least two predetermined radii that contains at least a predetermined minimum number of listings;

extracting, at the language model creation engine, all listings within the selected radius; and

creating, at the language model creation engine, a language model comprising all listings within the selected radius.

9. The method of claim 8, wherein the language model creation engine is part of a voice-driven system and the method further comprises the voice-driving search system:

receiving a location of a user;

selecting one of a plurality of language models stored in a language model database that covers an area including the location of the user;

receiving an audio input provided by the user; and

processing the audio input based at least in part upon the selected language model to determine a search request identified in the audio input.

10. The method of claim 9, wherein the steps performed by the voice-driven search system are performed by a network-based voice-driven search system.

11. The method of claim 9, wherein the steps performed by the voice-driven search system are performed by a mobile device-based voice-driven search system.

12. The method of claim 8 further comprising the language model creation engine excluding, from the geo-centric listing set, any listings that are listed in the local service area with an address and geo-coordinates outside of the local service area.

13. A non-transitory computer-readable medium comprising computer-executable instructions, for creating a geo-centric language model for use in performing a search for a user of a mobile-communications device, wherein the instructions, when executed by a processor, cause the processor to perform steps comprising:

determining a location of the mobile communications device;

determining a geographic center for a local service area, to be used in the search, based on the location of the mobile communications device, yielding a determined local service area;

computing a listing density, for the local service area, the listing density representing a density of searchable listings located in the determined local service area, yielding a computed listing density; and

determining the geo-centric language model to correspond to a region defined by the determined radius about the location of the mobile communications device.

14. The non-transitory computer-readable medium of claim 13, wherein the processor is part of a network-based voice-driven system and the steps of the method are performed by the processor for the network-based voice-driven system.

15. The non-transitory computer-readable medium of claim 13, wherein the processor is part of a mobile device-based voice-driven system and the steps of the method are performed by the processor for the device-based voice-driven system.

16. The non-transitory computer-readable medium of claim 13 further comprising computer-executable instructions that, when executed by the processor, cause the processor to perform the additional steps of:

receiving all the location;

determining the geo-centric language model, corresponding to the region defined by the determined radius about the location of the mobile communications device, including selecting one of a plurality of language models stored in a language model database that covers said region;

receiving an audio input provided by the user; and

processing the audio input based at least in part upon the determined language model to determine a search request identified in the audio input.

17. The non-transitory computer-readable medium of claim 16, wherein the computer-executable instructions in causing the processor to receive the location, causes the processor to receive the location as determined by a mobile device via at least one of cellular triangulation, WIFI triangulation, global positioning system (GPS), assisted GPS (A-GPS), area code, and user input, wherein the mobile device is remote to the processor.

18. The non-transitory computer-readable medium of claim 16, wherein the computer-executable instructions in causing the processor to receive the location, causes the processor to receive the location as determined by a mobile device via at least one of cellular triangulation, WIFI triangulation, GPS, A-GPS, area code, and user input, wherein the mobile device includes the processor.

19. The non-transitory computer-readable medium of claim 16 further comprising computer-executable instructions that, when executed by the processor, cause the processor to perform an additional step of excluding, from consideration in determining the geo-centric language model listing set, any national listings that are listed in the region but have an address or geo-coordinates outside of the region.

20. The non-transitory computer-readable medium of claim 16 further comprising computer-executable instructions that, when executed by the processor, cause the processor to perform an additional step of combining the geo-centric language model and a national language model using a count merging combination strategy or a language model union combination strategy.