Systems and methods for dynamically determining the attitude of a natural language speaker

• US 7,260,519 B2
• Filed: 03/13/2003
• Issued: 08/21/2007
• Est. Priority Date: 03/13/2003
• Status: Expired due to Fees

• Current Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.)
• Original Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.)
• Inventors: Polanyi, Livia, Zaenen, Annie, Baker, James D., Azara, Misty L.
• Primary Examiner(s): Hudspeth; David
• Assistant Examiner(s): ALBERTALLI, BRIAN LOUIS
• Application Number:
  US10/387,719
• Publication Number:
  US 20040186719A1
• Time in Patent Office: 1,622 Days
• Field of Search: None
• US Class Current: 704/9
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G10L 17/26 Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices

Abstract

Systems and methods for analyzing speech containing at least one lexical item, said analysis to determine an attitude of a speaker towards an entity, comprising determining at least one actual valence for the at least one lexical item by analyzing the at least one lexical item in context; determining the attitude based on the at least one actual valence; associating the speaker, the entity and the attitude; and wherein the at least one lexical item encodes attitude information about the entity.

Claims

1. A method for analyzing speech containing at least one lexical item, said analysis to determine an attitude of a speaker towards an entity, comprising:

determining at least one actual valence for the at least one lexical item by analyzing the at least one lexical item in context of said speech;

determining said attitude based on the at least one actual valence;

associating said speaker, said entity and said attitude;

wherein the at least one lexical item encodes attitude information about said entity;

determining at least one base valence for the at least one lexical item; and

wherein analyzing the at least one lexical item in context includes adjusting the at least one base valence based on a contextual valence shifter (CVS) selected from the group consisting of a presupposition CVS, a genre CVS and a tune.

2. The method of claim 1, wherein:

the at least one CVS is a presupposition CVS.

3. The method of claim 1, further comprising:

determining at least one additional CVS, where the additional CVS is one of;

a negator, an intensifier, a presupposition, a connector, a modal, reported speech/thought, an elaboration, a genre and a tune.

4. The method of claim 1, wherein:

the at least one CVS is a tune.

5. The method of claim 4 wherein:

the tune can scale the at least one base valence.

6. The method of claim 4 wherein:

the tune can indicate a different point of view and can neutralize or negate the at least one base valence.

7. The method of claim 3, wherein:

the negator CVS is a word or phrase that can negate the at least one base valence.

8. The method of claim 3, wherein:

the intensifier CVS is a word or phrase used for emphasis or de-emphasis that can strengthen or weaken the at least one base valence, respectively.

9. The method of claim 1, wherein:

the presupposition CVS can have a negative impact on the at least one base valence if the consequences of a not born-out expectation are negative and can have a positive impact on the at least one base valence if the consequences of the not born-out expectation are positive.

10. The method of claim 3, wherein:

the connector CVS is a word or phrase that indicates a different point of view and can neutralize or negate the at least one base valence.

11. The method of claim 3, wherein:

the modal CVS is a word or phrase that establishes a condition wherein the condition can neutralize or negate the at least one base valence.

12. The method of claim 3, wherein:

the reported speech/thought or other linguistic device CVS describes a context in which attitude is ascribed to a second entity rather than the speaker and can neutralize the at least one base valence with respect to the speaker.

13. The method of claim 12 wherein:

the reported speech/thought CVS does not neutralize the at least one base valence if the speaker adopts the attitude ascribed to the entity.

14. The method of claim 1, wherein:

the at least one CVS is a genre CVS.

15. The method of claim 4, wherein the tune includes a pitch accent, a phrase accent and a boundary tone.

16. The method of claim 14, wherein:

the genre CVS can adjust the base valence using knowledge of expected genre content.

17. The method of claim 16 wherein:

the genre can be a movie review.

18. The method of claim 1, wherein:

the at least one base valence represents positive or negative attitudinal information encoded in the lexical item.

19. The method of claim 1, wherein:

determining the at least one base valence is based on taking into account the domain in which the at least one lexical item occurs.

20. A method for analyzing speech containing at least one lexical item, said analysis to determine an attitude of a speaker towards an entity, comprising:

determining at least one base valence for said at least one lexical item;

determining at least one actual valence for the at least one lexical item by adjusting the at least one base valence based on at least one contextual valence shifter (CVS) selected from the group consisting of a presupposition CVS, a genre CVS and a tune;

determining said attitude based on the at least one actual valence;

associating said speaker, said entity and said attitude; and

wherein said at least one lexical item encodes attitude information about said entity.

21. The method of claim 20 further comprising determining at least one additional CVS, where the additional CVS is one of:

a negator, an intensifier, a presupposition, a connector, a modal, reported speech/thought, an elaboration, a genre, and a tune.

22. The method of claim 20, wherein:

the tune can scale the at least one base valence.

23. The method of claim 20, wherein:

the tune can indicate a different point of view and can neutralize or negate the at least one base valence.

24. The method of claim 21 wherein:

the negator CVS is a word or phrase that can negate the at least one base valence.

25. The method of claim 21 wherein:

the intensifier CVS is a word or phrase used for emphasis or de-emphasis that can strengthen or weaken the at least one base valence, respectively.

26. The method of claim 20, wherein:

the presupposition CVS can have a negative impact on the at least one base valence if the consequences of a not born-out expectation are negative and can have a positive impact on the at least one base valence if the consequences of the not born-out expectation are positive.

27. The method of claim 21 wherein:

the connector CVS is a word or phrase that indicates a different point of view and can neutralize or negate the at least one base valence.

28. The method of claim 21 wherein:

the modal CVS is a word or phrase that establishes a condition wherein the condition can neutralize or negate the at least one base valence.

29. The method of claim 21 wherein:

the reported speech/thought or other linguistic device CVS describes a context in which attitude is ascribed to a second entity rather than the speaker and can neutralize the at least one base valence with respect to the speaker.

30. The method of claim 29 wherein:

the reported speech/thought CVS does not neutralize the at least one base valence if the speaker adopts the attitude ascribed to the entity.

31. The method of claim 21 wherein:

the elaboration CVS adjusts the at least one base valence to reflect the valence of at least one elaborating lexical item that elaborates on the lexical item.

32. The method of claim 31 wherein:

a contradiction between the lexical item and the at least one elaborating lexical item can indicate irony.

33. The method of claim 20, wherein:

the genre CVS can adjust the base valence using knowledge of expected genre content.

34. The method of claim 33 wherein:

the genre can be a movie review.

35. The method of claim 20 wherein:

the at least one base valence represents positive or negative attitudinal information encoded in the lexical item.

36. The method of claim 20 wherein:

determining the at least one base valence is based on taking into account the domain in which the lexical item occurs.

37. The method of claim 20 wherein:

the at least one lexical item can be a contextual valence shifter.

38. The method of claim 20, wherein the tune includes a pitch accent, a phrase accent and a boundary tone.

39. A system for analyzing speech containing at least one lexical item, said analysis to determine an attitude of a speaker towards an entity, comprising:

a base valence calculator to determine at least one base valence of the at least one lexical item;

a contextual valence adjuster coupled to the base valence calculator, the contextual valence adjuster to adjust the at least one base valence based on at least one contextual valence shifter (CVS), selected from the group consisting of a presupposition CVS, a genre CVS and a tune;

an entity identifier to identity the entity;

an attitude mapper coupled to the entity identifier and the contextual valence adjuster, the attitude mapper to determine the attitude based on the at least one adjusted valence and to associate the speaker, the entity, and an attitude.

40. The system of claim 39 wherein:

the base valence calculator utilizes a lexical resource; and

wherein the lexical resource provides the at least one base valence of the at least one lexical item.

41. The system of claim 40 wherein:

the lexical resource uses domain information to determine the at least one base valence.

42. The system of claim 39 wherein:

the base valence calculator utilizes domain information to determine a domain for the at least one lexical item.

43. The system of claim 39 further comprising determining

at least one additional CVS, where the additional CVS is one of:

a negator, an intensifier, a presupposition, a connector, a modal, reported speech/thought, an elaboration, a genre, and a tune.

44. The system of claim 39, wherein:

the tune can scale the at least one base valence.

45. The system of claim 39, wherein:

the tune can indicate a different point of view and can neutralize or negate the at least one base valence.

46. The system of claim 43 wherein:

the negator CVS is a word or phrase that can negate the at least one base valence.

47. The system of claim 43 wherein:

the intensifier CVS is a word or phrase used for emphasis or de-emphasis that can strengthen or weaken the at least one base valence, respectively.

48. The system of claim 39, wherein:

the presupposition CVS can have a negative impact on the at least one base valence if the consequences of a not born-out expectation are negative and can have a positive impact on the at least one base valence if the consequences of the not born-out expectation are positive.

49. The system of claim 43 wherein:

the connector CVS is a word or phrase that indicates a different point of view and can neutralize or negate the at least one base valence.

50. The system of claim 43 wherein:

the modal CVS is a word or phrase that establishes a condition wherein the condition can neutralize or negate the at least one base valence.

51. The system of claim 43 wherein:

the reported speech/thought or other linguistic device CVS describes a context in which attitude is ascribed to a second entity rather than the speaker and can neutralize the at least one base valence with respect to the speaker.

52. The system of claim 51 wherein:

the reported speech/thought CVS does not neutralize the at least one base valence if the speaker adopts the attitude ascribed to the entity.

53. The system of claim 43 wherein:

the elaboration CVS adjusts the at least one base valence to reflect the valence of at least one elaborating lexical item that elaborates on the lexical item.

54. The system of claim 53 wherein:

a contradiction between the lexical item and the at least one elaborating lexical item can indicate irony.

55. The system of claim 39, wherein:

the genre CVS can adjust the base valence using knowledge of expected genre content.

56. The system of claim 55 wherein:

the genre can be a movie review.

57. The system of claim 39 wherein:

the at least one base valence represents positive or negative attitudinal information encoded in the lexical item.

58. The system of claim 39 wherein:

the at least one lexical item can be a contextual valence shifter.

59. The system of claim 39, wherein the tune includes a pitch accent, a phrase accent and a boundary tone.

60. A machine readable medium having instructions stored thereon that when executed by a processor cause a system to:

determine at least one actual valence for at least one lexical item by analyzing the at least one lexical item in context of a speech;

determine an attitude based on the at least one actual valence;

associate a speaker, an entity and the attitude;

wherein the at least one lexical item encodes attitude information about the entity;

determining at least one base valence for the at least one lexical item; and

wherein analyzing the at least one lexical item includes adjusting the at least one base valence based on a contextual valence shifter (CVS) selected from the group consisting of a presupposition CVS, a genre CVS and a tune.

61. The machine readable medium of claim 60, wherein the tune includes a pitch accent, a phrase accent and a boundary tone.

62. The machine readable medium of claim 60 further comprising determining:

at least one additional CVS, where the additional CVS is one of;

a negator, an intensifier, a presupposition, a connector, a modal, reported speech/thought, an elaboration, a genre, and a tune.

63. The machine readable medium of claim 60, wherein:

the tune can scale the at least one base valence.

64. The machine readable medium of claim 60, wherein:

the tune can indicate a different point of view and can neutralize or negate the at least one base valence.

65. The machine readable medium of claim 62 wherein:

the negator CVS is a word or phrase that can negate the at least one base valence.

66. The machine readable medium of claim 62 wherein:

the intensifier CVS is a word or phrase used for emphasis or de-emphasis that can strengthen or weaken the at least one base valence, respectively.

67. The machine readable medium of claim 60 wherein:

the presupposition CVS can have a negative impact on the at least one base valence if the consequences of a not born-out expectation are negative and can have a positive impact on the at least one base valence if the consequences of the not borne-out expectation are positive.

68. The machine readable medium of claim 62 wherein:

the connector CVS is a word or phrase that indicates a different point of view and can neutralize or negate the at least one base valence.

69. The machine readable medium of claim 62 wherein:

the modal CVS is a word or phrase that establishes a condition wherein the condition can neutralize or negate the at least one base valence.

70. The machine readable medium of claim 62 wherein:

the reported speech/thought or other linguistic device CVS describes a context in which attitude is ascribed to a second entity rather than the speaker and can neutralize the at least one base valence with respect to the speaker.

71. The machine readable medium of claim 70 wherein:

the reported speech/thought CVS does not neutralize the at least one base valence if the speaker adopts the attitude ascribed to the entity.

72. The machine readable medium of claim 62 wherein:

the elaboration CVS adjusts the at least one base valence to reflect the valence oft least one elaborating lexical item that elaborates on the lexical item.

73. The machine readable medium of claim 72 wherein:

a contradiction between the lexical item and the at least one elaborating lexical item can indicate irony.

74. The machine readable medium of claim 62 wherein:

the genre CVS can adjust the base valence using knowledge of expected genre content.

75. The machine readable medium of claim 74 wherein:

the genre can be a movie review.

76. The machine readable medium of claim 60 wherein:

the at least one base valence represents positive or negative attitudinal information encoded in the lexical item.

77. The machine readable medium of claim 60, wherein:

determining the at least one base valence is based on taking into account the domain in which the at least one lexical item occurs.

78. The machine readable medium of claim 60 wherein:

the at least one lexical item can be a contextual valence shifter.