Multi-level decision-analytic approach to failure and repair in human-computer interactions

• US 6,490,698 B1
• Filed: 06/04/1999
• Issued: 12/03/2002
• Est. Priority Date: 06/04/1999
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Paek, Timothy, Horvitz, Eric
• Primary Examiner(s): BADERMAN, SCOTT T
• Application Number:
  US09/326,043
• Time in Patent Office: 1,278 Days
• Field of Search: 714/46, 345/809, 704/200, 704/200.1, 704/246, 704/251, 704/257
• US Class Current: 714/46
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding

Abstract

A multi-level decision-analytic approach to failure and repair within computer-user communications is disclosed. In one embodiment, a computerized system for repairing communication failure within a computer-user interaction context includes a maintenance module, an intention module, and a conversation control subsystem. The maintenance module manages uncertainty regarding signal identification and channel fidelity. The intention module is supported by the maintenance module, and manages uncertainty about the recognition of user'"'"'s goals from signals. The conversation control subsystem surrounds both the modules, and manages the joint activity between the computer and the user, and one or more high-level events regarding the joint activity.

Claims

1. A computerized system for repairing communication failure within a computer-user interaction context comprising:

a maintenance module to manage uncertainty regarding signal identification and channel fidelity;

an intention module, supported by the maintenance module, to manage uncertainty regarding user goal recognition from signals; and

,

a conversation control subsystem surrounding the maintenance module and the intention module, to manage a joint activity and at least one high-level event regarding the joint activity.

2. The system of claim 1, wherein the maintenance module is to manage uncertainty of at least one of a channel level and a signal level.

3. The system of claim 1, wherein the intention module is to manage uncertainty at an intention level.

4. The system of claim 1, wherein the conversational control subsystem is to manage uncertainty at a conversation level.

5. The system of claim 1, wherein at least one of the conversation control subsystem, the intention module, and the maintenance module comprises a computer program executed by a processor from a computer-readable medium.

6. The system of claim 1, wherein at least one of the intention module and the maintenance module comprises a decision-based transition network comprising a plurality of nodes.

7. The system of claim 6, wherein at least the intention module comprises a decision-based transition network comprising a plurality of nodes, and the plurality of nodes of the network of the intention module comprise:

a request node;

an uptake node based on the request node; and

,

a side-sequence repair node based on the request node.

8. The system of claim 6, wherein the plurality of nodes comprise a plurality of transitional inputs.

9. The system of claim 8, wherein the plurality of transitional inputs comprise a selected transitional input.

10. The system of claim 9, wherein the selected transitional input comprises a selected transitional input based on an inference probability exceeding a threshold.

11. The system of claim 9, wherein the selected transitional input comprises a selected transitional input based on decision theoretic functions.

12. The system of claim 11, wherein the decision-theoretic functions attempt to minimize collaborative effort.

13. The system of claim 10, wherein the inference probability is determined by determining a probability distribution over leading variables of interest.

14. The system of claim 10, wherein the inference probability comprises an inference probability determined via a Bayesian network.

15. The system of claim 6, wherein at least one of the plurality of nodes comprises an instantiation of one of a group of possible nodes.

16. The system of claim 15, wherein the instantiation comprises a selected instantiation based on an inference probability exceeding a threshold.

17. The system of claim 16, wherein the inference probability comprises an inference probability determined via a Bayesian network.

18. The system of claim 15, wherein the instantiation comprises an instantiation for a dialog about one of:

an elaboration, and a repetition.

19. The system of claim 15, wherein the instantiation comprises an instantiation utilizing a value-of-information analysis.

20. The system of claim 19, wherein the value-of-information analysis comprises a value-of-information analysis based on utilities and costs affected by the at least one high-level event managed by the conversation control subsystem.

21. The system of claim 8, wherein each of the intention module and the maintenance module comprises a decision-based transition network comprising a plurality of nodes comprising a plurality of transitional inputs, and wherein the conversational control subsystem, based on a transitional input selected in the intention module, is to send evidence to the maintenance module to cause a transitional input to be fixed in the maintenance module.

22. The system of claim 1, wherein the conversation control subsystem is further to discern termination of the joint activity.

23. The system of claim 1, wherein the at least one high-level event comprises a number of questions asked.

24. A computerized system for repairing communication failure within a computer-user interaction context comprising:

a maintenance module having a first decision-based transition network comprising a first plurality of nodes having a first plurality of transitional inputs;

an intention module, supported by the maintenance module, having a second decision-based transition network comprising a second plurality of nodes having a second plurality of transitional inputs; and

,

a conversational control subsystem surrounding the maintenance module and the intention module.

25. The system of claim 24, wherein the conversational control subsystem is to manage a joint activity, at least one high-level event regarding the joint activity, and termination of the joint activity.

26. The system of claim 24, wherein at least one of the first and the second plurality of transitional inputs comprise a selected transitional input.

27. The system of claim 26, wherein the selected transitional input comprises a selected transitional input based on an inference probability exceeding a threshold, the inference probability determined via a Bayesian network.

28. The system of claim 24, wherein at least one of the first plurality of nodes and the second plurality of nodes comprises an instantiation of one of a group of possible nodes.

29. The system of claim 28, wherein the instantiation comprises a selected instantiation based on an inference probability exceeding a threshold, the inference probability determined via a Bayesian network.

30. The system of claim 28, wherein the instantiation comprises a selected instantiation utilizing a value-of-information analysis.

31. The system of claim 30, wherein the value-of-information analysis comprises a value-of-information analysis based on utilities and costs affected by at least one high-level event managed by the conversation control system.

32. The system of claim 24, wherein the conversational control subsystem, based on a transitional input selected in the intention module, is to send evidence to the maintenance module to cause a transitional input to be fixed in the maintenance module.

33. A computer-implemented method for repairing communication failure within a computer-user interaction context comprising:

managing uncertainty at a channel level and a signal level;

managing uncertainty at an intention level; and

,

managing uncertainty at a conversation level.

34. The method of claim 33, wherein managing uncertainty at a channel level and a signal level comprises utilizing a first decision-based transition network having a first plurality of nodes having a first plurality of transitional inputs, and managing uncertainty at an intention level comprises utilizing a second decision-based transition network having a second plurality of nodes having a second plurality of transitional inputs.

35. The method of claim 33, wherein managing uncertainty at a conversation level comprises managing a joint activity, at least one high-level event regarding the joint activity, and termination of the joint activity.

36. The method of claim 34, wherein at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network comprises selecting a selected transitional input.

37. The method of claim 36, wherein selecting a selected transitional input comprises:

determining an inference probability via a Bayesian network; and

,

selecting the selected transitional input based on the inference probability exceeding a threshold.

38. The method of claim 34, wherein at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network comprises instantiating an instantiation node.

39. The method of claim 38, wherein instantiating an instantiation node comprises:

determining an inference probability via a Bayesian network; and

,

instantiating the instantiation node based on the inference probability exceeding a threshold.

40. The method of claim 38, wherein the at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network further comprises utilizing a value-of-information analysis within the instantiation node.

41. The method of claim 40, wherein utilizing a value-of-information analysis comprises assessing costs and utilities affected by at least one managed high-level event managed.

42. The method of claim 34, further comprising:

selecting a second transitional input; and

,

fixing a first transitional input based on the second transitional input.

43. A machine-readable medium having instructions stored thereon for execution by a processor to cause performance of a method for repairing communication failure within a computer-user interaction context comprising:

utilizing a first decision-based transition network having a first plurality of nodes having a first plurality of transitional inputs to manage uncertainty at a channel level and a signal level;

utilizing a second decision-based transition network having a second plurality of nodes having a second plurality of transitional inputs to manage uncertainty at an intention level; and

,

managing uncertainty at a conversation level.

44. The medium of claim 43, wherein managing uncertainty at a conversation level comprises managing a joint activity, at least one high-level event regarding the joint activity, and termination of the joint activity.

45. The medium of claim 43, wherein at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network comprises selecting a selected transitional input.

46. The medium of claim 45, wherein selecting a selected transitional input comprises:

determining an inference probability via a Bayesian network; and

,

selecting the selected transitional input based on the inference probability exceeding a threshold.

47. The medium of claim 43, wherein at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network comprises instantiating an instantiation node.

48. The medium of claim 47, wherein instantiating an instantiation node comprises:

determining an inference probability via a Bayesian network; and

,

instantiating the instantiation node based on the inference probability exceeding a threshold.

49. The medium of claim 48, wherein the at least one of utilizing a first decision-based transition network and utilizing a second decision-based transition network further comprises utilizing a value-of-information analysis within the instantiation node.

50. The medium of claim 49, wherein utilizing a value-of-information analysis comprises assessing costs and utilities affected by at least one managed high-level event managed.

51. The medium of claim 43, further comprising:

selecting a second transitional input; and

,

fixing a first transitional input based on the second transitional input.