APPLICATION-DEPENDENT INFORMATION FOR RECOGNITION PROCESSING

• US 20100318359A1
• Filed: 06/10/2009
• Published: 12/16/2010
• Est. Priority Date: 06/10/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Herold, Keith C., Manocha, Sandeep, Hamaker, Jonathan E., Plumpe, Michael D., Odell, Julian James
• Granted Patent: US 8,442,826 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams

Abstract

Architecture for integrating application-dependent information into a constraints component at deployment time or when available. In terms of a general grammar, the constraints component can include or be a general grammar that comprises application-independent information and is structured in such a way that application-dependent information can be integrated into the general grammar without loss of fidelity. The general grammar includes a probability space and reserves a section of the probability space for the integration of application-dependent information. An integration component integrates the application-dependent information into the reserved section of the probability space for recognition processing. The application-dependent information is integrated into the reserved section of the probability space at deployment time or when available. The general grammar is structured to support the integration and improve the overall system.

Claims

1. A computer-implemented recognition system, comprising:

a constraints component for parsing input data, the constraints component interfaces to a probability space; and

an integration component for integrating application-dependent information into the constraints component for the recognition processing.

2. The system of claim 1, wherein the constraints component, which includes a general grammar, reserves a section of the probability space for integration of the application-dependent information.

3. The system of claim 1, wherein the application-dependent information is integrated directly into a reserved section of the probability space post-deployment.

4. The system of claim 1, wherein the application-dependent information is integrated into a reserved section of the probability space when available.

5. The system of claim 1, wherein the constraints component is structured to support the integration.

6. The system of claim 1, wherein the application-dependent information comprises at least one of a language model, an acoustic model, or other speech recognition system.

7. The system of claim 1, wherein the application-dependent information applies to a specific instance of use.

8. The system of claim 1, wherein the application-dependent information accumulates over multiple uses.

9. The system of claim 1, wherein the application-dependent information is optional and the constraints component remains functional when the application-dependent information is unavailable.

10. A computer-implemented recognition system, comprising:

a constraint components which includes a general grammar that contains application-independent information and a defined probability space, a section of the probability space reserved for integration of application-dependent information; and

an integration component for integrating the application-dependent information directly into the reserved section of the probability space at time of use.

11. The system of claim 10, wherein the application-dependent information is structured to optimally support the integration.

12. The system of claim 10, wherein the reserved section of the probability space is fixed or varies dynamically based on user needs.

13. The system of claim 10, wherein the application-dependent information comprises embedded slot grammars specific to a user from an n-gram language model.

14. The system of claim 10, wherein the application-dependent information applies to a specific instance of use or accumulates over multiple uses.

15. A computer-implemented method of recognition processing of data, comprising:

defining a probability space within a set of constraints for recognition processing of data;

reserving a section of the probability space for integrating application-dependent information; and

integrating the application-dependent information into the reserved section of the probability space as a sub-set of the constraints.

16. The method of claim 15, further comprising integrating the application-dependent information into the reserved section of the probability space after deployment.

17. The method of claim 15, further comprising integrating the application-dependent information into the reserved section of the probability space when available.

18. The method of claim 15, further comprising structuring the application-dependent information to support the integration.

19. The method of claim 15, further comprising mining at least one of word documents, emails, or other data to create the set of constraints, which include a general grammar.

20. The method of claim 15, further comprising:

utilizing at least one of user specific information, a specific scenario, a specific environment, or other information specific to a user interaction as application-dependent information; and

biasing specific categories of the application-dependent information with various weights based on a likelihood that the application-dependent information will be used.