Method and system for automatically detecting morphemes in a task classification system using lattices

• US 7,440,897 B1
• Filed: 05/27/2006
• Issued: 10/21/2008
• Est. Priority Date: 11/05/1999
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Gorin, Allen Louis, Petrovska-Delacretaz, Dijana, Riccardi, Giuseppe, Wright, Jeremy Huntley
• Primary Examiner(s): Azad, Abul K.
• Application Number:
  US11/420,738
• Time in Patent Office: 878 Days
• Field of Search: 704/254, 704/249
• US Class Current: 704/254
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

In an embodiment, a lattice of phone strings in an input communication of a user may be recognized, wherein the lattice may represent a distribution over the phone strings. Morphemes in the input communication of the user may be detected using the recognized lattice. Task-type classification decisions may be made based on the detected morphemes in the input communication of the user.

Claims

1. A method of detecting morphemes using lattices in an automated task classification system which operates on one or more task objectives of a user, comprising:

recognizing a lattice of phone strings in an input communication of the user, the lattice representing a distribution over the phone strings;

detecting morphemes in the input communication of the user using the recognized lattice; and

making task-type classification decisions based on the detected morphemes in the input communication of the user.

2. The method of claim 1, wherein the morphemes include at least one of verbal speech or non-verbal speech.

3. The method of claim 2, wherein the non-verbal speech includes use of at least one of gestures, body movements, head movements, non-responses, text, keyboard entries, keypad entries, mouse clicks, DTMF codes, pointers, stylus, cable set-top box entries, graphical user interface entries or touchscreen entries.

4. The method of claim 2, wherein each of the verbal and non-verbal speech is directed to a respective one of the one or more task objectives and each of the verbal and non-verbal speech is labeled with the respective one of the one or more task objectives to which it is directed.

5. The method of claim 1, wherein the morphemes are expressed in multimodal form.

6. The method of claim 1, wherein the input communication of the user is derived from the verbal and non-verbal speech and an environment of the user.

7. The method of claim 1, wherein the morphemes in the input communication of the user are derived from actions of the user, including a focus of attention of the user.

8. The method of claim 1, further comprising prompting the user to provide a feedback response including additional information with respect to an initial input communication of the user.

9. The method of claim 1, further comprising prompting the user to provide a feedback response that includes a confirmation with respect to at least one of the task objectives determined in the classification decision.

10. The method of claim 1, wherein the input communication of the user is routed based on a respective one of the classification decisions.

11. The method of claim 1, wherein the classification decisions and corresponding input communications of the user are collected for automated learning purposes.

12. The method of claim 1, wherein a relationship between the detected morphemes and the one or more task objectives includes a salience measure of one of the detected morphemes with respect to a specified one of a plurality of predetermined task objectives.

13. The method of claim 12, wherein the salience measure is represented as a conditional probability of one of the one or more task objectives being requested given an appearance of the morpheme in the input communication, the conditional probability being a highest value in a distribution of the conditional probabilities over the plurality of predetermined task objectives.

14. The method of claim 12, wherein each of the plurality of detected morphemes has a salience measure exceeding a predetermined threshold.

15. The method of claim 1, wherein a relationship between the detected morphemes and a predetermined set of task objectives includes a mutual information measure within a language of the morphemes.

16. The method of claim 15, wherein each of the detected morphemes has a mutual information measure exceeding a predetermined threshold.

17. The method 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 method of claim 1, further comprising:

generating morphemes by clustering selected ones of semantically and syntactically similar salient sub-morphemes from training speech; and

storing the generated morphemes in a database which is used to detect morphemes in the recognized lattice.

19. A method of task classification which operates on one or more task objectives of a user, comprising:

selecting salient phone-phrases from training speech;

generating morphemes by clustering selected ones of the salient phone-phrases which are semantically and syntactically similar;

recognizing a lattice of phone strings in an input communication of the user, the lattice representing a distribution over the phone strings;

detecting morphemes in the input communication of the user using the recognized lattice; and

making task-type classification decisions based on the detected morphemes in the input communication of the user.

20. An automated task classification system that detects morphemes using lattices and operates on one or more task objectives of a user, comprising:

a recognizer that recognizes a lattice of phone strings in an input communication of the user, the lattice representing a distribution over the phone strings;

a morpheme detector that detects morphemes in the input communication of the user using the recognized lattice; and

a task classification processor that makes task-type classification decisions based on the detected morphemes in the input communication of the user.

21. The automated task classification system of claim 20, wherein the morphemes include at least one of verbal speech or non-verbal speech.

22. The automated task classification system of claim 21, wherein the non-verbal speech includes the use of at least one of gestures, body movements, head movements, non-responses, text, keyboard entries, keypad entries, mouse clicks, DTMF codes, pointers, stylus, cable set-top box entries, graphical user interface entries or touchscreen entries.

23. The automated task classification system of claim 20, wherein the morphemes are expressed in multimodal form.

24. The automated task classification system of claim 20, wherein the input communication of the user is derived from verbal speech and non-verbal speech and an environment of the user.

25. The automated task classification system of claim 20, wherein the morphemes in the input communication of the user are derived from actions of the user, including a focus of attention of the user.

26. The automated task classification system of claim 20, wherein the task classification processor is arranged to prompt the user to provide a feedback response including additional information with respect to an initial input communication of the user.

27. The automated task classification system of claim 20, wherein the task classification processor is arranged to prompt the user to provide a feedback response that includes a confirmation with respect to at least one of a plurality of task objectives determined in the classification decision.

28. The automated task classification system of claim 20, wherein the classification decisions and corresponding user input communications are collected for automated learning purposes.