MANAGING USER INTERACTION FOR INPUT UNDERSTANDING DETERMINATIONS

• US 20160196499A1
• Filed: 01/07/2015
• Published: 07/07/2016
• Est. Priority Date: 01/07/2015
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Khan, Omar Zia, Sarikaya, Ruhi
• Granted Patent: US 10,572,810 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 3/167 Audio in a user interface, e.g. using voice commands for navigating, audio feedback
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Examples of the present disclosure improve decision-making for input understanding to assist in determining how to best respond to a user input. A received input is analyzed using an input recognition component, input understanding component and input context component. Potential response options are determined. If uncertainty exists with respect to responding to the received input, an uncertainty value and a cost of misclassification are generated for the potential response options to assist in making a decision as to how to best respond to the received input. The uncertainty value is determined for a potential response and parameters associated with the potential response and the cost of misclassification is a cost associated with pursuing a potential response if the potential response turns out to be incorrect. A response is selected to transmit to a user based on analyzing the generated uncertainty value and the generated cost of misclassification for the potential responses.

Claims

1. A system comprising:

a memory; and

at least one processor connected with the memory and configured to execute processing for components of an input understanding platform including;

a user interaction mediation component to;

evaluate potential response options for an input when the system is uncertain of a response to the input, wherein the user interaction mediation component evaluates the potential response options based on;

an uncertainty value associated with a potential response option and parameters of the potential response option, and

a cost of misclassification of pursuing a potential response option if the potential response option turns out to be incorrect, and

select a response from the potential response options to transmit to the user based on evaluating the uncertainty value and the cost of misclassification for the potential response options.

2. The system according to claim 1, wherein the input understanding platform further comprising:

an input recognition component to receive, process and translate the input for recognition, and determine a confidence value associated with recognition of the input;

an input understanding component to analyze the input for understanding including determining alternative understandings of the analyzed input and determining a confidence value associated with the alternative understandings of the analyzed input; and

an input context component to evaluate context of the analyzed input including determining possible response options based on evaluation of the context of the analyzed input and determining a confidence value for each potential response option based on evaluation of the parameters of each potential response option, and

wherein the user interaction mediation component aggregates confidence level values received from the input recognition component, the input understanding component and the input context component to determine the uncertainty value.

3. The system according to claim 2, wherein the input is a voice input received by the system, the input recognition component is a speech recognition component, the input understanding component is a spoken language understanding component, and the input context component is a dialog policy component.

4. The system according to claim 1, wherein the user interaction mediation component determines modes of clarifying the received input if the selected response is a request for further clarification of the input based on a context of the selected, generates a cost value associated with each of the determined modes of clarifying, selects a mode of clarifying the input based on evaluation of the generated cost value for the determined modes of clarifying, and identifies parameters associated with the mode of clarifying for the user to respond to based on analyzing the input using knowledge information managed by the system.

5. The system according to claim 1, wherein the user interaction mediation component determines, for the response, whether to execute an action or request further clarification by one of:

analyzing discriminative classifiers for the cost of misclassification and the uncertainty value for a potential response, and modeling probability data based on a combination of the cost of misclassification and the uncertainty value for a potential response.

6. The system according to claim 1, wherein the cost of misclassification is a cost value estimating an impact if a potential response option turns out to be incorrect and if the parameters associated with the potential response option turn out to be incorrect, and wherein the cost of misclassification determined based on at least one of subjective measures and empirical data.

7. A computer-implemented method comprising:

analyzing received input using an input recognition component, input understanding component and input context component;

generating, for potential responses to the received input, an uncertainty value determined based on a potential response and parameters associated with the potential response and a cost of misclassification associated with pursuing a potential response if the potential response turns out to be incorrect, when uncertainty exists with respect to a response to the received input; and

selecting a response to transmit to a user based on analyzing the generated cost of misclassification and the generated uncertainty value for the potential responses.

8. The computer-implemented method according to claim 7, further comprising:

comparing potential responses by evaluating the generated cost of misclassification and the generated uncertainty value for each potential response, and

wherein the selecting selects the response based on the comparing.

9. The computer-implemented method according to claim 7,

wherein the generating of the uncertainty value for a potential response is based on aggregating confidence values obtained during the analyzing of the received input from each of the input recognition component, the input understanding component and the input context component.

10. The computer-implemented method according to claim 7, wherein the selecting of the response is determined based at least one of:

analyzed discriminative classifiers for the generated cost of misclassification and the generated uncertainty value for each of the potential responses, and probability data combining the cost of misclassification and the uncertainty value for a potential response.

11. The computer-implemented method according to claim 7, wherein the response is one of an action executed in response to the received input, a request for further clarification of the received input, and a response including implicit confirmation of a received input.

12. The computer-implemented method according to claim 7, further comprising:

determining modes of clarifying the received input if the selected response is a request for further clarification of the received input,

generating a cost value associated with each of the determined modes of clarifying, and selecting a mode of clarifying the received input to provide the user based on evaluation of the generated cost value for the determined modes of clarifying.

13. The computer-implemented method according to claim 12, wherein the selecting of the mode of clarifying further comprises identifying parameters associated with the mode of clarifying for the user to respond to based on analyzing the received input using knowledge data stored in at least one knowledge store.

14. The computer-implemented method according to claim 7, further comprising:

transmitting the response to the user;

analyzing another received input from the user including applying knowledge data associated with a previously received input to the another received input;

generating, for potential responses to the another received input, an uncertainty value determined based on a potential response to the another received input and parameters associated with the potential response and a cost of misclassification for a case where the potential response to the another received input turns out to be incorrect, when uncertainty exists with respect to a response to the another received input; and

selecting another response to transmit to the user based on analyzing the generated cost of misclassification and the generated uncertainty value for the potential responses to the another received input.

15. The computer-implemented method according to claim 7, wherein the received input is at least one of a voice input, a gesture input, a text input and a handwritten input.

16. A computer-readable storage device containing instructions, that when executed on at least one processor, causing the processor to execute a process comprising:

analyzing received input using an input understanding platform, wherein confidence values are determined upon components of the input understanding platform evaluating the received input;

aggregating the determined confidence values components of the input understanding platform to determine a level of uncertainty with respect to a potential response to the received input;

determining whether to request further clarification from a user based on evaluating the level of uncertainty for the potential response and a cost of misclassification associated with pursuing a potential response option if the potential response option turns out to be incorrect; and

selecting a response to transmit to a user based on the determining.

17. The computer-readable storage device according to claim 16, wherein the process executed by the processor comprising:

determining modes of clarifying the received input if the selected response is a request for further clarification of the received input,

generating a cost value associated with each of the determined modes of clarifying, and

selecting a mode of clarifying the received input to provide the user based on evaluation of the generated cost value for each of the determined modes of clarifying.

18. The computer-readable storage device according to claim 17, wherein the selecting of the mode of clarifying further comprises identifying parameters associated with the mode of clarifying for the user to respond to based on analyzing the received input using stored knowledge data.

19. The computer-readable storage device according to claim 16, wherein the determining executed by the processor determines whether to execute an action or request further clarification by at least one of analyzing discriminative classifiers for the cost of misclassification and the level of uncertainty for the potential response, and analyzing combined data based on the cost of misclassification and the level of uncertainty for the potential response.

20. The computer-readable storage device according to claim 16, wherein the selecting executed by the processor selects a response from a plurality of potential response options by comparing the level of uncertainty and the cost of misclassification for the potential responses options.