Intelligent user assistance facility

• US 6,021,403 A
• Filed: 07/19/1996
• Issued: 02/01/2000
• Est. Priority Date: 07/19/1996
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Rommelse, Jacobus A., Horvitz, Eric, Heckerman, David E., Hovel, David O., Klein, Adrian C., Breese, John S., Shaw, Gregory L., Hobson, Samuel D.
• Primary Examiner(s): Hafiz, Tariq R.
• Assistant Examiner(s): Rhodes, Jason W.
• Application Number:
  US08/684,003
• Time in Patent Office: 1,292 Days
• Field of Search: 706/11, 706/45, 706/46, 706/47, 706/52, 706/61
• US Class Current: 706/45
• CPC Class Codes: G06F 9/453 Help systems
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

An event composing and monitoring system that allows high-level events to be created from combinations of low-level events. An event specification tool, contained in the system, allows for rapidly developing a general event processor that creates high-level events from combinations of user actions. An event system, in combination with an inference system, monitors and infers, for various purposes, about several classes of events including: current program context; state of key data structures; user input sequences, including actions with a mouse-controlled cursor while interacting with a graphical user interface; words typed in free-text help queries; visual user information, such as gaze and gesture information; and user speech information. Additionally, an intelligent user interface is provided by constructing a reasoning model that computes probability of alternative user intentions, goals or information needs through analyzing information regarding program state, and that user'"'"'s actions and free-text query words. Specifically, the interface monitors user interaction with a program and probabilistically reasons to sense that a user may need assistance in using a particular feature or to accomplish a specific task. This interface accepts a free-text help query from the user and combines the inference analysis of user actions and the program state with an inference analysis of the query. The inference system, using an updateable user profile, continually checks for user competencies and, based on such competencies, changes assistance that is offered.

Claims

1. In a computer having a processor, and a memory, connected to the processor, having computer-executable instructions stored therein, a method for providing a user help facility for use in conjunction with a program then executing in the computer, the computer also having an output device for providing a user interface through which a user can interact with the program, the method comprising the steps of:

during execution of the program, performing a probabilistic assessment of a current user interaction with the interface so as to yield a numeric probability distribution across a plurality of different available help topics such that a separate probability measure is generated for each of said help topics, the interaction evincing an intention by the user to accomplish a specific task through the program, hence defining a plurality of separate probability measures wherein each measure represents a probability of whether the user, given the interaction, then needs information about a specific corresponding one of the topics for accomplishing the task, wherein the interaction has both textual and non-textual components, the textual component constituting at least one word which the user has entered through the interface, as a free-text query, and the non-textual component constituting either an affirmative user action with the interface or an absence of said affirmative action during a predefined interval of time that has occurred since an immediately prior user interaction with the interface, wherein the affirmative action comprises a manipulation of a portion of the interface by the user or a verbal, visual or gestural response by the user to the interface;

selecting, based on the plurality of probability measures in the distribution, a particular one of the help topics as being most relevant to the user interaction; and

providing a predefined help message, associated with the one help topic, to the user through the output device.

2. The method in claim 1 wherein the probabilistic assessment performing step comprises the step of undertaking the probabilistic assessment in response to a current context of the program.

3. The method in claim 2 wherein the context comprises current states of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

4. The method in claim 3 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

5. The method in claim 3 wherein the visual response comprises eye focus or gaze.

6. The method in claim 3 wherein the gestural response comprises a facial gesture.

7. The method in claim 3 wherein the verbal response comprises speech or an utterance produced by the user.

8. The method in claim 2 wherein the probabilistic assessment performing step comprises the steps of:

determining that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously providing without prompting from the user, the predefined help message to the user thus providing autonomous help.

9. The method in claim 8 further comprising the step, while rendering said autonomous help to the user, of acquiring information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

10. The method in claim 2 wherein the probabilistic assessment performing step comprises the step of modifying, for use in a subsequent inference analysis and based on a time interval that has elapsed since the current user interaction occurred, probability measures to be used in a probabilistic assessment of a next successive user interaction.

11. The method in claim 2 wherein the probabilistic assessment performing step comprises the step of undertaking the probabilistic assessment through Bayesian analysis.

12. The method in claim 11 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

13. The method in claim 2 wherein the help message providing step comprises the step of generating a social user interface, on the output device, through which the user can interact with the user help facility.

14. The method in claim 13 wherein the social user interface comprises an animated character.

15. The method in claim 2 wherein the probabilistic assessment performing step comprises the steps of:

undertaking a first probabilistic assessment of the textual component to yield a first probability distribution and a first set of corresponding ones of the help topics;

undertaking, independent of the first probabilistic assessment, a second probabilistic assessment of the non-textual component, in conjunction with the current context, to yield a second probability distribution and a second set of corresponding ones of the help topics; and

selecting, from the first and second sets of the help topics, the particular one of the help topics.

16. The method in claim 15 further comprising the step of selecting, as being most relevant to the user interaction, the particular one help topic, from the first and second sets of help topics, as that having a maximum corresponding probability value.

17. The method in claim 16 wherein the probabilistic assessment performing step further comprises the steps of:

weighting individual probability measures associated with each of the help topics contained in the first and second sets of help topics by first and second predefined numeric weights, respectively, so as to yield first and second sets of weighted probability measures, wherein the first and second weights are different numerical values; and

selecting the particular one help topic as that which is then associated with a maximum value in the first and second sets of weighted probability measures.

18. The method in claim 16 wherein the visual response comprises eye focus or gaze.

19. The method in claim 16 wherein the gestural response comprises a facial gesture.

20. The method in claim 16 wherein the verbal response comprises speech or an utterance produced by the user.

21. The method in claim 16 further comprising the step of performing the first and second probabilistic assessments through Bayesian analysis.

22. The method in claim 21 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

23. The method in claim 15 wherein the context comprises current states of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

24. The method in claim 23 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

25. The method in claim 15 wherein the help message providing step comprises the step of generating a social user interface, on the output device, through which the user can interact with the user help facility.

26. The method in claim 25 wherein the social user interface comprises an animated character.

27. The method in claim 2 wherein the probabilistic assessment performing step comprises the steps of:

monitoring user activity with the interface so as to generate a stream of atomic user events, wherein each such atomic user event represents a separate and distinct user action, said user action constituting any one of a group of events comprising;

a separate non-textual manipulation of a portion of the interface;

the verbal, visual or gestural response to the interface;

or

an entry, through the interface, of said one word as part of the free-text query;

generating, through an event processor and in response to the atomic events and time of occurrence of each of the atomic events, a succession of modeled events, wherein each one of the modeled events represents, according to a predetermined composition rule, an occurrence of a corresponding predefined set of at least one of the atomic events;

inferring, through an inference system and in response to each of the modeled events and the current context, the probability distribution across the plurality of said help topics so as to yield the plurality of probability measures for the help topics; and

selecting, as being most relevant to the user interaction, the particular one help topic as that which is then associated with a maximum measure in the probability distribution.

28. The method in claim 27 wherein the computer further comprises a user profile stored in the memory, the user profile storing predefined ones of the modeled events so as to indicate user competencies, wherein the probabilistic assessment performing step comprises the step of, if one of the modeled events is one of the predefined events, storing the one predefined event in the user profile;

and wherein the probability distribution inferring step is also responsive to the user profile for inferring the probability distribution across the plurality of said help topics.

29. The method in claim 28 wherein the probability distribution inferring step further comprises the step of inferring the probability distribution across the plurality of said help topics, additionally in response to a history of help topics previously presented to the user.

30. The method in claim 29 wherein the probability distribution inferring step further comprises the step, if no user profile or said help topics history exists for the user, of inferring the probability distribution across the plurality of said help topics through the inference system and, in lieu of the user profile and the help topics history, in response to competency information manually supplied to the interface by the user.

31. The method in claim 27 wherein the probability distribution inferring step comprises the steps of:

determining, through the inference system, that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously providing, without prompting from the user, the predefined help message to the user thus providing autonomous help.

32. The method in claim 31 further comprising the step, while rendering said autonomous help to the user, of acquiring information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

33. The method in claim 27 wherein the probability distribution inferring step comprises the step of modifying the probability measures generated as a result of the occurrence of said one modeled event, for use in a subsequent inference analysis, based on a time interval that has elapsed since ones of the atomic events that constitute said one modeled event occurred.

34. The method in claim 27 wherein the probability distribution inferring step comprises the step of performing the first and second probabilistic assessments through Bayesian analysis.

35. The method in claim 34 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

36. The method in claim 27 wherein the help message providing step comprises the step of generating a social user interface, on the output device, through which the user can interact with the user help facility.

37. The method in claim 36 wherein the social user interface comprises an animated character.

38. The method in claim 27 wherein the probability distribution inferring step comprises the steps of:

undertaking a first inference analysis for both the current context and first ones of said modeled events so as to yield a first probability distribution and a first set of corresponding ones of the help topics, said first ones of said modeled events resulting from the user actions associated with a separate non-textual manipulation of a portion of the interface, or a verbal, visual or gestural response to the interface;

undertaking a second inference analysis, independent of the first inference analysis, for second ones of the modeled events so as to yield a second probability distribution and a second set of corresponding ones of the help topics, said second ones of the modeled events resulting from user actions associated with entry of said word as part of a free-text query; and

selecting, from the first and second sets of the help topics, the particular one of the help topics.

39. The method in claim 38 further comprising the step of selecting, as being most relevant to th e user interaction, the particular one help topic, from the first and second sets of help topics, as that having a maximum probability value.

40. The method in claim 39 wherein the probabilistic distribution inferring step comprises the steps of:

weighting individual probability measures associated with each of the help topics contained in the first and second sets of help topics by first and second predefined numeric weights, respectively, so as to yield first and second sets of weighted probability measures, wherein the first and second weights are different numerical values; and

selecting the particular one help topic as that which is then associated with a maximum value in the first and second sets of weighted probability measures.

41. The method in claim 39 wherein the context comprises state of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

42. The method in claim 41 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

43. The method in claim 39 wherein the visual response comprises eye focus or gaze.

44. The method in claim 39 wherein the gestural response comprises a facial gesture.

45. The method in claim 39 wherein the verbal response comprises speech or an utterance produced by the user.

46. The method in claim 39 wherein the probabilistic distribution inferring step comprises the step of performing the first and second probabilistic assessments through Bayesian analysis.

47. The method in claim 46 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

48. The method in claim 27 wherein the computer further comprises a user profile stored in the memory, the user profile storing predefined ones of the model ed events so as to indicate user competencies, wherein the probability distribution inferring step comprises the step, if one of the modeled events is one of the p redefined events, of storing the one predefined event in the user profile;

a nd wherein the probabilistic distribution inferring step is also responsive to the user profile, for inferring the probability distribution across the plurality of said help topics.

49. The method in claim 48 wherein the probability distribution inferring step comprises the step of inferring the probability distribution across the plurality of said help topics, additionally in response to a history of help topics previously presented to the user.

50. The method in claim 49 wherein the probability distribution inferring step comprises the step, if no user profile or said help topics history exists for the user, of inferring the probability distribution across the plurality of said help topics through the inference system and, in lieu of the user profile and the help topics history, in response to competency information manually supplied to the interface by the user.

51. The method in claim 27 wherein the probabilistic distribution inferring step comprises the steps of:

determining, through the inference system, that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously providing, without prompting the user, the predefined help message to the user thus providing the autonomous help.

52. The method in claim 51 wherein the probabilistic distribution inferring step comprises the step, while rendering said autonomous help to the user, of acquiring information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

53. The method in claim 27 wherein the probabilistic distribution inferring step comprises the step of modifying the probability measures generated as a result of the occurrence of said one modeled event, for use in a subsequent inference analysis, based on a time interval that has elapsed since ones of the atomic events that constitute said one modeled event occurred.

54. The method in claim 27 wherein the probabilistic distribution inferring step comprises the step of performing the first and second probabilistic assessments through Bayesian analysis.

55. The method in claim 54 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

56. The method in claim 27 wherein the help message providing step comprises the step of generating a social user interface, on the output device, through which the user can interact with the user help facility.

57. The method in claim 56 wherein the social user interface comprises an animated character.

58. The method in claim 2 wherein the computer further comprises a speech recognizer for recognizing the verbal response supplied by the user during the interaction with the interface, the speech recognizer having a language model, containing entries collectively mapping current user needs to corresponding probabilities of occurrence of given words or utterances then being produced by the user, and adjusting its operation in response to the language model, and the method further comprises the step of modifying the probabilities of occurrence of said given words or utterances in the language model in response to the probabilistic assessment of the user interaction such that probabilities of occurrence of ones of verbal responses that are more likely to occur, given current needs of the user, are increased, while probabilities of occurrence of other ones of the verbal responses that are less likely to occur, given the current needs of the user, are decreased.

59. The method in claim 58 further comprising the step of modifying the probabilities of occurrence in the language model in response to the current context of the program.

60. A computer readable medium having computer executable instructions stored therein for performing the steps of claim 1.

61. In a computer, apparatus for providing a user help facility for use in conjunction with a program then executing in the computer, the computer having an output device for providing a user interface through which a user can interact with the program, the apparatus comprising:

a processor; and

a memory, connected to the processor, having computer-executable instructions stored therein; and

wherein the processor, in response to the stored instructions;

during execution of the program, performs a probabilistic assessment of a current user interaction with the interface so as to yield a numeric probability distribution across a plurality of different available help topics such that a separate probability measure is generated for each of said help topics, the interaction evincing an intention by the user to accomplish a specific task through the program, hence defining a plurality of separate probability measures wherein each measure represents a probability of whether the user, given the interaction, then needs information about a specific corresponding one of the topics for accomplishing the task, wherein the interaction has both textual and non-textual components, the textual component constituting at least one word which the user has entered through the interface, as a free-text query, and the non-textual component constituting either an affirmative user action with the interface or an absence of said affirmative action during a predefined interval of time that has occurred since an immediately prior user interaction with the interface, wherein the affirmative action comprises a manipulation of a portion of the interface by the user or a verbal, visual or gestural response by the user to the interface;

selects, based on the plurality of probability measures in the distribution, a particular one of the help topics as being most relevant to the user interaction; and

provides a predefined help message, associated with the one help topic, to the user through the output device.

62. The apparatus in claim 61 wherein the processor, in response to the stored instructions, performs the probabilistic assessment in response to a current context of the program.

63. The apparatus in claim 62 wherein the context comprises current states of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

64. The apparatus in claim 63 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

65. The apparatus in claim 63 wherein the visual response comprises eye focus or gaze.

66. The apparatus in claim 63 wherein the gestural response comprises a facial gesture.

67. The apparatus in claim 63 wherein the verbal response comprises speech or an utterance produced by the user.

68. The apparatus in claim 62 wherein the processor, in response to the stored instructions:

determines that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously provides, without prompting from the user, the predefined help message to the user thus providing autonomous help.

69. The apparatus in claim 68 wherein the processor, in response to the stored instructions and while rendering said autonomous help to the user, acquires information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

70. The apparatus in claim 62 wherein the processor, in response to the stored instructions, modifies, for use in a subsequent inference analysis and based on a time interval that has elapsed since the current user interaction occurred, probability measures to be used in a probabilistic assessment of a next successive user interaction.

71. The apparatus in claim 62 wherein the processor, in response to the stored instructions, performs the probabilistic assessment through Bayesian analysis.

72. The apparatus in claim 71 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

73. The apparatus in claim 62 wherein the processor, in response to the stored instructions, generates a social user interface, on the output device, through which the user can interact with the user help facility.

74. The apparatus in claim 73 wherein the social user interface comprises an animated character.

75. The apparatus in claim 62 wherein the processor, in response to the stored instructions:

undertakes a first probabilistic assessment of the textual component to yield a first probability distribution and a first set of corresponding ones of the help topics;

undertakes, independent of the first probabilistic assessment, a second probabilistic assessment of the non-textual component, in conjunction with the current context, to yield a second probability distribution and a second set of corresponding ones of the help topics; and

selects, from the first and second sets of the help topics, the particular one of the help topics.

76. The apparatus in claim 75 wherein the processor, in response to the stored instructions and through the selecting operation, selects, as being most relevant to the user interaction, the particular one help topic, from the first and second sets of help topics, as that having a maximum corresponding probability value.

77. The apparatus in claim 76 wherein the processor, in response to the stored instructions:

weights individual probability measures associated with each of the help topics contained in the first and second sets of help topics by first and second predefined numeric weights, respectively, so as to yield first and second sets of weighted probability measures, wherein the first and second weights are different numerical values; and

selects the particular one help topic as that which is then associated with a maximum value in the first and second sets of weighted probability measures.

78. The apparatus in claim 76 wherein the visual response comprises eye focus or gaze.

79. The apparatus in claim 76 wherein the gestural response comprises a facial gesture.

80. The apparatus in claim 76 wherein the verbal response comprises speech or an utterance produced by the user.

81. The apparatus in claim 76 wherein the processor, in response to the stored instructions, performs the first and second probabilistic assessments through Bayesian analysis.

82. The apparatus in claim 81 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

83. The apparatus in claim 76 wherein the processor, in response to the stored instructions, generates a social user interface, on the output device, through which the user can interact with the user help facility.

84. The apparatus in claim 83 wherein the social user interface comprises an animated character.

85. The apparatus in claim 75 wherein the context comprises current states of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

86. The apparatus in claim 85 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

87. The apparatus in claim 62 wherein the processor, in response to the stored instructions:

monitors user activity with the interface so as to generate a stream of atomic user events, wherein each such atomic user event represents a separate and distinct user action, said user action constituting any one of a group of events comprising;

a separate non-textual manipulation of a portion of the interface;

the verbal, visual or gestural response to the interface;

or

an entry, through the interface, of said one word as part of the free-text query;

generates, through an event processor and in response to the atomic events and time of occurrence of each of the atomic events, a succession of modeled events, wherein each one of the modeled events represents, according to a predetermined composition rule, an occurrence of a corresponding predefined set of at least one of the atomic events;

infers, through an inference system and in response to each of the modeled events and the current context, the probability distribution across the plurality of said help topics so as to yield the plurality of probability measures for the help topics; and

selects, as being most relevant to the user interaction, the particular one help topic as that which is then associated with a maximum measure in the probability distribution.

88. The apparatus in claim 87 wherein the apparatus further comprises a user profile stored in the memory, the user profile storing predefined ones of the modeled events so as to indicate user competencies, wherein the processor, in response to the stored instructions,

if one of the modeled events is one of the predefined events, stores the one predefined event in the user profile;

and

infers, through the inference system and additionally in response to the user profile, the probability distribution across the plurality of said help topics.

89. The apparatus in claim 88 wherein the inference system, in response to the stored instructions, infers the probability distribution across the plurality of said help topics, additionally in response to a history of help topics previously presented to the user.

90. The apparatus in claim 89 wherein processor, in response to the stored instruction and if no user profile or said help topics history exists for the user, infers the probability distribution across the plurality of said help topics through the inference system and, in lieu of the user profile and the help topics history, in response to competency information manually supplied to the interface by the user.

91. The apparatus in claim 87 wherein the processor, in response to the stored instructions:

determines, through the inference system, that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously provides, without prompting from the user, the predefined help message to the user thus providing autonomous help.

92. The apparatus in claim 91 wherein the processor, in response to the stored instructions and while rendering said autonomous help to the user, acquires information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

93. The apparatus in claim 87 wherein the processor, in response to the stored instructions, modifies the probability measures generated as a result of the occurrence of said one modeled event, for use in a subsequent inference analysis, based on a time interval that has elapsed since ones of the atomic events that constitute said one modeled event occurred.

94. The apparatus in claim 87 wherein the processor, in response to the stored instructions, performs the first and second probabilistic assessments through Bayesian analysis.

95. The apparatus in claim 94 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

96. The apparatus in claim 87 wherein the processor, in response to the stored instructions, generates a social user interface, on the output device, through which the user can interact with the user help facility.

97. The apparatus in claim 96 wherein the social user interface comprises an animated character.

98. The apparatus in claim 87 wherein the processor, in response to the stored instructions, and through the inferring operation:

undertakes a first inference analysis for both the current context and first ones of said modeled events so as to yield a first probability distribution and a first set of corresponding ones of the help topics, said first ones of said modeled events resulting from the user actions associated with a separate non-textual manipulation of a portion of the interface, or a verbal, visual or gestural response to the interface;

undertakes a second inference analysis, independent of the first inference analysis, for second ones of the modeled events so as to yield a second probability distribution and a second set of corresponding ones of the help topics, said second ones of the modeled events resulting from user actions associated with entry of said word as part of a free-text query; and

selects, from the first and second sets of the help topics, the particular one of the help topics.

99. The apparatus in claim 98 wherein the processor, in response to the stored instructions and through the selecting operation, selects, as being most relevant to the user interaction, the particular one help topic, from the first and second sets of help topics, as that having a maximum probability value.

100. The apparatus in claim 99 wherein the processor, in response to the stored instructions:

weights individual probability measures associated with each of the help topics contained in the first and second sets of help topics by first and second predefined numeric weights, respectively, so as to yield first and second sets of weighted probability measures, wherein the first and second weights are different numerical values; and

selects the particular one help topic as that which is then associated with a maximum value in the first and second sets of weighted probability measures.

101. The apparatus in claim 99 wherein the context comprises state of program data structures, and the non-textual component comprises a sequence of manipulations of portions of the interface provided, as input, by the user.

102. The apparatus in claim 101 wherein the user input sequences comprise user manipulations of a mouse-controlled cursor depicted in the interface.

103. The apparatus in claim 99 wherein the visual response comprises eye focus or gaze.

104. The apparatus in claim 99 wherein the gestural response comprises a facial gesture.

105. The apparatus in claim 99 wherein the verbal response comprises speech or an utterance produced by the user.

106. The apparatus in claim 99 wherein the processor, in response to the stored instructions, performs the first and second probabilistic assessments through Bayesian analysis.

107. The apparatus in claim 106 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

108. The apparatus in claim 99 wherein the processor, in response to the stored instructions, generates a social user interface, on the output device, through which the user can interact with the user help facility.

109. The apparatus in claim 108 wherein the social user interface comprises an animated character.

110. The apparatus in claim 87 wherein the apparatus further comprises a user profile stored in the memory, the user profile storing predefined ones of the modeled events so as to indicate user competencies, wherein the processor, in response to the stored instructions,

if one of the modeled events is one of the predefined events, stores the one predefined event in the user profile;

and

infers, through the inference system and additionally in response to the user profile, the probability distribution across the plurality of said help topics.

111. The apparatus in claim 110 wherein the inference system, in response to the stored instructions, infers the probability distribution across the plurality of said help topics, additionally in response to a history of help topics previously presented to the user.

112. The apparatus in claim 111 wherein processor, in response to the stored instruction and if no user profile or said help topics history exists for the user, infers the probability distribution across the plurality of said help topics through the inference system and, in lieu of the user profile and the help topics history, in response to competency information manually supplied to the interface by the user.

113. The apparatus in claim 87 wherein the processor, in response to the stored instructions:

determines, through the inference system, that a probability of the user needing assistance exceeds a predefined threshold value so as to define an autonomous help condition; and

in response to the autonomous help condition, autonomously provides, without prompting the user, the predefined help message to the user thus providing the autonomous help.

114. The apparatus in claim 113 wherein the processor, in response to the stored instructions and while rendering said autonomous help to the user, acquires information from the user where an expected value of the information, as defined by a preset utility function, exceeds a current cost associated with obtaining the information from the user.

115. The apparatus in claim 87 wherein the processor, in response to the stored instructions, modifies the probability measures generated as a result of the occurrence of said one modeled event, for use in a subsequent inference analysis, based on a time interval that has elapsed since ones of the atomic events that constitute said one modeled event occurred.

116. The apparatus in claim 87 wherein the processor, in response to the stored instructions, performs the first and second probabilistic assessments through Bayesian analysis.

117. The apparatus in claim 116 wherein the Bayesian analysis comprises either a Bayesian network or a Bayesian influence diagram.

118. The apparatus in claim 62 further comprising a speech recognizer for recognizing the verbal response supplied by the user during the interaction with the interface, the speech recognizer having a language model, containing entries collectively mapping current user needs to corresponding probabilities of occurrence of given words or utterances then being produced by the user, and adjusting its operation in response to the language model, and the processor, in response to the stored instructions, modifies the probabilities of occurrence of said given words or utterances in the language model in response to the probabilistic assessment of the user interaction such that probabilities of occurrence of ones of verbal responses that are more likely to occur, given current needs of the user, are increased, while probabilities of occurrence of other ones of the verbal responses that are less likely to occur, given the current needs of the user, are decreased.

119. The apparatus in claim 118 wherein processor, in response to the stored instructions, modifies the probabilities of occurrence in the language model in response to the current context of the program.