Using multiple modality input to feedback context for natural language understanding

Patent metadata

Field	Value
Publication number	US-11264023-B2
Application number	US-201916419105-A
Country	US
Kind code	B2
Filing date	May 22, 2019
Priority date	Aug 29, 2011
Publication date	Mar 1, 2022
Grant date	Mar 1, 2022

Abstract

Input context for a statistical dialog manager may be provided. Upon receiving a spoken query from a user, the query may be categorized according to at least one context clue. The spoken query may then be converted to text according to a statistical dialog manager associated with the category of the query and a response to the spoken query may be provided to the user.

First claim

We claim:

1. A system for generating statistical dialog managers for recognizing speech, the system comprising:
at least one processor; and
a memory storage operatively coupled to the at least one processor, the memory storage storing instructions that when executed by the at least one processor cause the system to perform a set of operations comprising:
receiving user data for generating one or more statistical dialog managers;
identifying a plurality of categories associated with the user data, wherein the plurality of categories include a first category and a second category;
based on the received user data and the identified plurality of categories, building a plurality of statistical dialog managers, wherein the plurality of statistical dialog managers includes a first statistical dialog manager for the first category and a second statistical dialog manager for the second category; and
converting a spoken query received in a first input field to text according to at least one of the plurality of statistical dialog managers, wherein the spoken query corresponds to the first category, wherein the first category is identified based on a user-inputted value for a second input field corresponding to the second category, and wherein the second category is based on the second statistical dialog manger statistically weight terms belonging to the second category.

2. The system of claim 1 , wherein the operations further comprise:
receiving additional user data;
based on the additional user data, updating at least one of the generated plurality of statistical dialog managers.

3. The system of claim 1 , wherein building the plurality of statistical dialog managers comprises upweighting terms in the received user data.

4. The system of claim 1 , wherein the user data includes a timestamp and a user id.

5. The system of claim 1 , wherein the user data includes data collected from an interactive form.

6. The system of claim 1 , wherein the user data includes a plurality of messages exchanged between users.

7. The system of claim 1 , wherein the user data includes a plurality of session logs.

8. The system of claim 1 , wherein the operations further comprise:
receiving a spoken user query;
categorizing the spoken user query as the first category; and
based on categorizing the spoken query as the first category, converting the spoken query to text according to the first statistical dialog manager.

9. The system of claim 8 , wherein categorizing the spoken query is based at least one of the following: a domain name, an input field name, a profile associated with the user, a previous query, and a user-inputted value associated with a second input field.

10. The system of claim 8 , wherein the spoken user query is received from a user device and the statistical dialog manager executes on a server communicatively coupled to the user device.

11. A method for generating statistical dialog managers for recognizing speech, the method comprising:
receiving user data for generating one or more statistical dialog managers;
identifying a plurality of categories associated with the user data, wherein the plurality of categories include a first category and a second category;
based on the received user data and the identified plurality of categories, building a plurality of statistical dialog managers, wherein the plurality of statistical dialog managers includes a first statistical dialog manager for the first category and a second statistical dialog manager for the second category; and
converting a spoken query received in a first input field to text according to at least one of the plurality of statistical dialog managers, wherein the spoken query corresponds to the first category, wherein the first category is identified based on a user-inputted value for a second input field corresponding to the second category, and wherein the second category is based on the second statistical dialog manger statistically weight terms belonging to the second category.

12. The method of claim 11 , further comprising:
receiving additional user data;
based on the additional user data, updating at least one of the generated plurality of statistical dialog managers.

13. The method of claim 11 , wherein building the plurality of statistical dialog managers comprises upweighting terms in the received user data.

14. The method of claim 11 , wherein the user data includes a timestamp and a user id.

15. The method of claim 11 , wherein the user data includes data collected from an interactive form.

16. The method of claim 11 , wherein the user data includes a plurality of session logs.

17. The method of claim 11 , further comprising:
receiving a spoken user query;
categorizing the spoken user query as the first category; and
based on categorizing the spoken query as the first category, converting the spoken query to text according to the first statistical dialog manager.

18. A computer-implemented method for recognizing speech, the computer-implemented method comprising:
receiving user data for generating one or more statistical dialog managers;
identifying a plurality of categories associated with the user data, wherein the plurality of categories include a first category and a first subcategory for the first category;
based on the received user data and the identified plurality of categories, building a plurality of statistical dialog managers, wherein the plurality of statistical dialog managers includes a first statistical dialog manager for the first category and a second statistical dialog manager for the first subcategory;
receiving a spoken user query;
categorizing the spoken user query as the first category; and
based on categorizing the spoken query as the first category, converting the spoken query to text according to the first statistical dialog manager, wherein the spoken query corresponds to the first category, wherein the first category is identified based on a user-inputted value for a second input field corresponding to the second category, and wherein the second category is based on the second statistical dialog manager with statistically weight terms belonging to the second category.

19. The method of claim 18 , wherein building the plurality of statistical dialog managers comprises upweighting terms in the received user data.

20. The method of claim 18 , wherein the user data includes data collected from an interactive form.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Bodell Michael
• Bain John
• Chambers Robert
• Cross Karen M
• Kim Michael
• Gedge Nick
• Penn Daniel Frederick
• Patel Kunal
• Tecot Edward Mark
• Waltmunson Jeremy C

Classifications

• G10L15/22 Primary

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• G10L2015/227

  of the speaker; Human-factor methodology · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title

• G10L2015/228

  of application context · CPC title