Automated task classification system

• US 7,590,224 B1
• Filed: 10/31/2000
• Issued: 09/15/2009
• Est. Priority Date: 09/15/1995
• Status: Expired due to Fees

• Current Assignee: AT&T Corporation (AT&T, Inc.)
• Original Assignee: AT&T Intellectual Property II LP (AT&T, Inc.)
• Inventors: Scherer, James B., Wilpon, Jay Gordon, Gorin, Allen Louis, Parker, Barry Allen
• Primary Examiner(s): Kuntz, Curtis
• Assistant Examiner(s): Phan, Joseph T
• Application Number:
  US09/699,495
• Time in Patent Office: 3,241 Days
• Field of Search: 704/246, 704/251, 704/270, 704/270.1, 379/88.01
• US Class Current: 379/88.13
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06Q 30/06 Buying, selling or leasing transactions
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding

Abstract

The invention concerns an automated task classification system that operates on a task objective of a user. The system may include a meaningful phrase generator that generates a plurality of meaningful phrases from a set of verbal and non-verbal speech. Each of the meaningful phrases may be generated based on one of a predetermined set of the task objectives. A recognizer may recognize at least one of the generated meaningful phrases in an input communication of the user and a task classifier may make a classification decision in response to the recognized meaningful phrases relating to one of the set of predetermined task objectives.

Claims

1. An automated task classification system that operates on a task objective of a user, through a natural language dialog with the user in which system prompts are not ordered in a menu, the system comprising:

a recognizer that spots at least one of a plurality of meaningful phrases in substantially simultaneous user natural language verbal and non-verbal input, wherein the natural language verbal and non-verbal input each convey different information and are associated with a coordinated message that achieves an appropriate response, each of the plurality of meaningful phrases having an association with at least one of a predetermined set of task objectives, and the predetermined set of task objectives based, at least partly, on a salience measure of one of the plurality of meaningful phrases to a specified one of the predetermined task objectives, wherein the salience measure is represented as a conditional probability of the task objective being requested given an appearance of one of the plurality of meaningful phrases in the input communication, the conditional probability being a highest value in a distribution of conditional probabilities over the set of predetermined task objectives; and

a task classifier that makes a classification decision based, at least partly, on the spotted at least one of the plurality of meaningful phrases.

2. The automated task classification system of claim 1, wherein the meaningful phrases are expressed in a multimodal form.

3. The automated task classification system of claim 2, wherein the multimodal form includes inputs from at least one channel.

4. The automated task classification system of claim 1, wherein the meaningful phrases in the user'"'"'s input communication received by the recognizer are derived from the user'"'"'s actions.

5. The automated task classification system of claim 1, further comprising a dialog module that enters into a dialog with the user to obtain a feedback response from the user.

6. The automated task classification system of claim 5, wherein the dialog module prompts the user to provide a feedback response that includes additional information with respect to the user'"'"'s initial input communication.

7. The automated task classification system of claim 5, wherein the dialog module prompts the user to provide a feedback response that includes confirmation with respect to at least one of the set of task objectives determined in the classification decision.

8. The automated task classification system of claim 1, wherein the task classifier routes the input communication based on the classification decision.

9. The automated task classification system of claim 8, wherein the task objective is performed after the input communication is routed by the task classifier.

10. The automated task classification system of claim 1, wherein the system is used for customer care purposes.

11. The automated task classification system of claim 1, wherein the classification decision and the corresponding input communication of the user are collected by the system for automated learning purposes.

12. The automated task classification system of claim 1, wherein each of the plurality of meaningful phrases has a salience measure exceeding a predetermined threshold.

13. The automated task classification system of claim 1, wherein the association between the meaningful phrases and the predetermined set of task objectives is based, at last partly, on a measure of commonality within a language of the meaningful phrases.

14. The automated task classification system of claim 13, wherein the measure of commonality is a mutual information measure.

15. The automated task classification system of claim 14, wherein each of the plurality of meaningful phrases has a mutual information measure exceeding a predetermined threshold.

16. The automated task classification system of claim 1, wherein the task classifier makes the classification decision using a confidence function.

17. The automated task classification system of claim 1, wherein the input communication from the user represents a request for at least one of the set of predetermined task objectives.

18. The automated task classification system of claim 1, wherein the input communication is responsive to a query of a form “

How may I help you?”

.

19. The automated task classification system of claim 1, wherein each of the natural language verbal input and the non-verbal input are directed to one of the set of predetermined task objectives and each of the natural language verbal input and the non-verbal input is labeled with the one task objective to which it is directed.

39. The automated task classification system of claim 1, through a natural language dialog with the user in which system prompts are not ordered in a menu, the system further comprising:

an interpretation module configured to apply a confidence function based on a probabilistic relation between the spotted at least one of the plurality of meaningful phrases in the input communication of the user and the at least one of the predetermined set of task objectives, wherein

the task classifier makes the classification decision based, at least partly on, a result of the applied confidence function.

20. An automated routing system that automatically routes a user'"'"'s request based on an automated task classification decision, through a natural language dialog with the user in which system prompts are not ordered in a menu, the system comprising:

a recognizer that spots at least one of the plurality of meaningful phrases in substantially simultaneous user natural language verbal and non-verbal input, wherein the natural language verbal and non-verbal input each convey different information and are associated with a coordinated message that achieves an appropriate response, each of the plurality of meaningful phrases having an association with at least one of a predetermined set of task objectives, and the predetermined set of task objectives based, at least partly, on a salience measure of one of the plurality of meaningful phrases to a specified one of the predetermined task objectives, wherein the salience measure is represented as a conditional probability of the task objective being requested given an appearance of one of the plurality of meaningful phrases in the input communication, the conditional probability being a highest value in a distribution of conditional probabilities over the set of predetermined task objectives;

a task classifier that makes a classification decision based, at least partly, on the spotted at least one of the plurality of meaningful phrases; and

a task router that routes the user'"'"'s request in order to perform at least one of the task objectives based on the classification decision.

21. The automated routing system of claim 20, wherein the meaningful phrases are expressed in multimodal form.

22. The automated routing system of claim 21, wherein the multimodal form includes inputs from at least one channel.

23. The automated routing system of claim 20, wherein the meaningful phrases in the user'"'"'s input communication received by the recognizer are derived from the user'"'"'s actions.

24. The automated routing system of claim 20, further comprising a dialog module that enters into a dialog with the user to obtain a feedback response from the user.

25. The automated routing system of claim 24, wherein the dialog module prompts the user to provide a feedback response that includes additional information with respect to the user'"'"'s request.

27. The automated routing system of claim 24, wherein if the task classifier cannot make a classification decision after the dialog is conducted with the user, the router routes the user'"'"'s request to a human for assistance.

28. The automated routing system of claim 27, wherein the task objective is performed after the user'"'"'s request is routed.

29. The automated routing system of claim 20, wherein the system is used for customer care purposes.

30. The automated routing system of claim 20, wherein the classification decision and the corresponding user request are collected by the system for automated learning purposes.

31. The automated routing system of claim 20, wherein each of the plurality of meaningful phrases has a salience measure exceeding a predetermined threshold.

32. The automated routing system of claim 20, wherein the association between the plurality of meaningful phrases and the predetermined set of task objectives is based, at least partly, on a measure of commonality with a language of the plurality of meaningful phrases.

33. The automated routing system of claim 32, wherein the measure of commonality is a mutual information measure.

34. The automated routing system of claim 33, wherein each of the plurality of meaningful phrases has a mutual information measure exceeding a predetermined threshold.

35. The automated routing system of claim 20, wherein the task classifier makes the classification decision using a confidence function.

36. The automated routing system of claim 20, wherein the user'"'"'s request represents a request for at least one of the set of predetermined task objectives.

37. The automated routing system of claim 20, wherein the user'"'"'s request is responsive to a query of a form “

How may I help you?”

.

38. The automated routing system of claim 20, wherein each of the natural language verbal input and the non-verbal input are directed to one of the set of predetermined task objectives and each of the natural language verbal input and the non-verbal input being labeled with the one task objective to which it is directed.

40. The automated routing system of claim 20, further comprising

an interpretation module configured to apply a confidence function based on a probabilistic relation between the spotted at least one of the plurality of meaningful phrases in the user'"'"'s request and the at least one of the predetermined set of task objectives, wherein

the task classifier makes the classification decision based, at least partly, on a result of the applied confidence function.

26. The automated routing system of clam 24, wherein the dialog module prompts the user to provide a feedback response that includes confirmation with respect to at least one of the set of task objectives determined in the classification decision.