Method and an apparatus to disambiguate requests

• US 9,131,050 B2
• Filed: 04/25/2012
• Issued: 09/08/2015
• Est. Priority Date: 01/29/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Agarwal, Rajeev, Ardman, David, Master, Muneeb, Mauro, David Andrew, Raman, Vijay R., Ulug, Amy E., Valli, Zulfikar
• Primary Examiner(s): DEANE JR, WILLIAM J
• Application Number:
  US13/456,144
• Publication Number:
  US 20120209594A1
• Time in Patent Office: 1,231 Days
• Field of Search: 379/218.01, 379/218.02, 37926501-26514, 37926601-2661
• US Class Current: 1/1
• CPC Class Codes: G06F 16/33 Querying
• CPC Class Codes: H04M 2201/10 Logic circuits
• CPC Class Codes: H04M 2201/39 using speech synthesis
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 3/4931 Directory assistance systems

Abstract

A method and an apparatus to disambiguate requests are presented. In one embodiment, the method includes receiving a request for information from a user. Then data is retrieved from a back-end database in response to the request. Based on a predetermined configuration of a disambiguation system and the data retrieved, the ambiguity within the request is dynamically resolved.

Claims

1. A computer-implemented method comprising:

receiving voice input comprising a request for information from a user;

retrieving data from a database within a system in response to the request;

determining that multiple distinctive entries of data are found to match the request;

engaging the user in a verbal dialog to determine which of the multiple distinctive entries satisfy the user'"'"'s request for information, the verbal dialog further comprising;

dynamically generating a prompt to request clarification information from the user;

rendering the prompt as speech and presenting the verbal prompt to the user; and

responsive to the prompt, receiving subsequent voice input from the user;

analyzing the subsequent voice input to retrieve the clarification information;

dynamically resolving ambiguity by

removing from the response to the request one or more of the multiple distinctive entries that are not consistent with the clarification information, and

applying one or more policies from a set of predetermined policies, wherein each policy of the set of predetermined policies is activated via a policy configuration file; and

determining that the ambiguity is resolved.

2. The method of claim 1, further comprising:

determining a number entries of data that are found to match the request;

responsive to determining that the number of entries of data that are found to match the request exceeds a predefined threshold, generating the prompt to be an open-ended question.

3. The method of claim 1, wherein the prompt is a close-ended question if the data retrieved includes a number of entries below a predefined threshold.

4. The method of claim 1, wherein determining that multiple distinctive entries of data are found to match the request further comprises:

executing pre-loaded rules on the retrieved data;

determining whether a confirmation or disambiguation dialog is required; and

responsive to determining that a confirmation dialog is required, directing the voice browser to a confirmation dialog rendering template.

5. The method of claim 1, further comprising:

selecting one or more entries of the multiple distinctive entries based on the analysis of the subsequent voice input; and

presenting the one or more selected entries to the user.

6. The method of claim 1, the verbal dialog further comprising:

generating a grammar based on a template; and

analyzing the subsequent voice input using the generated grammar.

7. The method of claim 1, further comprising:

responsive to analyzing the subsequent voice input, dynamically generating a second prompt to request further clarification information from the user;

rendering the second prompt as speech and presenting the verbal second prompt to the user;

responsive to the second prompt, receiving second subsequent voice input from the user;

analyzing the second subsequent voice input;

selecting one or more entries of the multiple distinctive entries based on the analysis of the second subsequent voice input; and

presenting the one or more selected entries to the user.

8. The method of claim 7, further comprising:

determining whether selecting one or more entries of the multiple distinctive entries satisfies a configured predefined condition; and

responsive to determining that the configured predefined condition is not satisfied, rendering an error message.

9. An apparatus comprising:

a user interface operable to allow an integrator of a directory assistance system to define a configuration of the directory assistance system;

a voice interface device operable to receive a spoken request for information from a user;

a disambiguation engine operable to dynamically resolve ambiguity within the user request, and operatively coupled to the voice interface device for retrieving data from a database in response to the request;

responsive to determining that multiple distinctive entries of data are found to match the request, the disambiguation engine is operable to interact with the user via the voice interface device to determine which of the multiple distinctive entries satisfy the user'"'"'s request for information, the interaction with the user further comprising;

dynamically generating a prompt to request clarification information from the user;

rendering the prompt as speech and presenting the verbal prompt to the user through the voice interface device; and

receiving subsequent voice input from the user via the voice interface device in response to the verbal prompt;

analyzing the subsequent voice input to interpret the clarification information;

dynamically resolving ambiguity by

removing from the response to the request one or more of the multiple distinctive entries that are not consistent with the clarification information, and

applying one or more policies from a set of predetermined policies, wherein each policy of the set of predetermined policies is activated via a policy configuration file; and

determining that the ambiguity is resolved.

10. The apparatus of claim 9, wherein the disambiguation engine comprises:

a prompt template processor to generate the prompt using a first predefined template.

11. The apparatus of claim 10, wherein the voice interface device comprises a voice browser to aurally present the prompt to a user.

12. The apparatus of claim 11, wherein the voice interface device is configured to receive a spoken response;

and further comprising;

a grammar template processor to generate a grammar using a second predefined template used by the disambiguation engine to recognize the response.

13. The apparatus of claim 9, wherein the user interface comprises a second user interface control to allow the integrator to modify the configuration.

14. A system comprising:

a database; and

a directory assistance system operatively coupled to the database, the directory assistance system comprising;

a voice interface device from which to receive a spoken request for information from a user;

a database access module to retrieve data from the database in response to the request; and

a configurable disambiguation engine operable to dynamically resolve ambiguity within the request based on a predetermined configuration of the directory assistance system and the data retrieved,

wherein the disambiguation engine comprises;

a rules processor operable to determine which of multiple distinctive entries satisfy the user'"'"'s request for information;

a controller operable to determine whether clarification information is needed to select among the multiple distinctive entries;

a prompt template processor operable to dynamically generate a first prompt to request clarification information from the user;

wherein the voice interface device presents the prompt verbally to the user through the voice interface device;

wherein the voice interface device is configured to receive subsequent voice input from the user in response to the verbal prompt; and

an ambiguity detector operable to;

analyze the subsequent voice input to interpret the clarification information;

dynamically resolve ambiguity by

removing from the response to the request one or more of the multiple distinctive entries that are not consistent with the clarification information, and

applying one or more policies from a set of predetermined policies, wherein each policy of the set of predetermined policies is activated via a policy configuration file; and

determine that the ambiguity is resolved.

15. The directory assistance system of claim 14, wherein the disambiguation engine is operable to select a portion of the data retrieved based on a response from the user to the first prompt.

16. The directory assistance system of claim 14, wherein the prompt template processor is operable, if the ambiguity remains, to generate a second prompt based on at least one of:

the response to the first prompt,

the portion of the data retrieved, or

a second rule within the configuration.

17. The directory assistance system of claim 14, wherein the voice interface device comprises a voice browser to aurally present the first prompt to the user.

18. A non-transitory machine-readable medium having stored thereon instructions which, when executed by a processor, cause the processor to perform operations comprising:

receiving voice input comprising a request for information from a user;

retrieving data from a database within a system in response to the request;

determining that multiple distinctive entries of data are found to match the request;

dynamically generating a prompt to request clarification information from the user;

rendering the prompt as speech and presenting a verbal prompt to the user;

responsive to the prompt, receiving subsequent voice input from the user;

analyzing the subsequent voice input to retrieve the clarification information;

dynamically resolving ambiguity by

removing from the response to the request one or more of the multiple distinctive entries that are not consistent with the clarification information, and

applying one or more policies from a set of predetermined policies, wherein each policy of the set of predetermined policies is activated via a policy configuration file; and

determining that the ambiguity is resolved.

19. The non-transitory machine-readable medium of claim 18, wherein the operations further comprise:

creating a user interface to allow an integrator to configure a directory assistance system and to modify the configuration.