AUTOMATED SENTENCE PLANNING IN A TASK CLASSIFICATION SYSTEM

• US 20090222267A1
• Filed: 02/26/2009
• Published: 09/03/2009
• Est. Priority Date: 03/14/2001
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Walker, Marilyn A., Rambow, Owen Christopher, Rogati, Monica
• Granted Patent: US 8,019,610 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 13/027 Concept to speech synthesisers; Generation of natural phrases from machine-based concepts generation of parameters for speech synthesis out of text G10L13/08
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 2203/355 Interactive dialogue design tools, features or methods
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

The invention relates to a task classification system (900) that interacts with a user. The task classification system (900) may include a recognizer (920) that may recognize symbols in the user'"'"'s input communication, and a natural language understanding unit (900) that may determine whether the user'"'"'s input communication can be understood. If the user'"'"'s input communication can be understood, the natural language understanding unit (930) may generate understanding data. The system may also include a communicative goal generator that may generate communicative goals based on the symbols recognized by the recognizer (920) and understanding data from the natural language understanding unit (930). The generated communicative goals may be related to information needed to be obtained from the user. The system may further include a sentence planning unit (120) that may automatically plan one or more sentences based on the communicative goals generated by the communicative goal generator with at least one of the sentences plans being output to the user.

Claims

1. A spoken dialog system comprising:

a processor;

a module configured to control the processor to generate a plurality of sentence plans based on a plurality of communicative goals associated with a user input communication;

a module configured to control the processor to rank the plurality of sentence plans generated by the sentence plan generator, wherein at least one of the ranked sentence plans is output to the user, and wherein the user input communication and the output at least one ranked sentence plan comprise a single dialog turn in a spoken dialog of the user input communication and the output at least one ranked sentence plan.

2. The spoken dialogue system of claim 1, further comprising:

a recognizer that recognizes symbolism in the user input communication.

3. The spoken dialogue system of claim 2, further comprising:

a natural language understanding unit that generates understanding data from the user input communication.

4. The spoken dialogue system of claim 3, further comprising:

a communicative goal generator that generates the plurality of communicative goals based on the symbols recognized by the recognizer and understanding data from the natural language understanding unit, the generated plurality of communicative goals being related to information needed to be obtained from the user and based on the user input communication.

5. The spoken dialogue system of claim 3, wherein the user input communication immediately preceeds the generation of the plurality of communicative goals.

6. The spoken dialogue system of claim 1, further comprising:

a training database that includes a set of learned rules, wherein the module configured to control the processor to rank the plurality of sentence plans generated by the sentence planning generator uses the set of learned rules.

7. The spoken dialogue system of claim 1, further comprising:

a discourse history database that includes interaction information related to a set of interactions between the user and the spoken dialogue system, wherein the module configured to control the processor to generate the plurality of sentence plans generates the plurality of sentence plans using the interaction information.

8. The spoken dialogue system of claim 1, further comprising:

a dialogue manager that conducts the dialogue with the user if a natural language dialogue unit does not understand recognized symbols in the user input communication.

9. The spoken dialogue system of claim 8, wherein the dialogue manager determines whether all of the communicative goals have been met.

10. The spoken dialogue system of claim 9, further comprising:

a task classification processor that routes information obtained from system interactions with the user for task completion if the dialogue manager determines that all of the communicative goals have been met.

11. The spoken dialogue system of claim 9, further comprising:

a task classification processor that processes any tasks associated with the information obtained from system interaction with the user if the dialogue manager determines that all of the communicative goals have been met.

12. The spoken dialogue system of claim 1, wherein a communicative goal generator generates communicative goals that include confirming information previously obtained from the user.

13. The spoken dialogue system of claim 1, further comprising a text-to-speech converter that converts at least one of the ranked sentence plans from text-to-speech.

14. The spoken dialogue system of claim 1, wherein the spoken dialogue system is used in one of a customer care system, a reservation system, parts ordering system, navigation system, information gathering system and information retrieval system.

18. The method of claim 13, further comprising interacting with the user to obtain task completion if a dialogue management determines that all of the communicative goals have not been met.

19. The method of claim 13, further comprising outputting the at least one of the ranked sentence plans to the user via a text-to-speech converter.

15. A computer-implemented method for engaging in a dialogue with the user, the method comprising:

generating a plurality of sentence plans based on a plurality of communicative goals associated with a user input communication; and

ranking the plurality of sentence plans generated by the sentence plan generator, wherein at least on of the ranked sentence plans is output to the user, and wherein the user input communication and the output at least one ranked sentence plan comprise a single dialog turn in a spoken dialog of the user input communication and the output at least one ranked sentence plan.

16. The method of claim 15, wherein the sentence plan ranker selects the highest ranked sentence plan.

17. The method of claim 15, further comprising ranking the plurality of sentence plans using a learned set of rules from a training database that includes a set of learned rules.

20. A tangible computer readable medium storing instructions for controlling a spoken dialog system, the instructions comprising:

generating a plurality of sentence plans based on a plurality of communicative goals associated with a user input communication; and

ranking the plurality of sentence plans generated by the sentence plan generator, wherein at least on of the ranked sentence plans is output to the user, and wherein the user input communication and the output at least one ranked sentence plan comprise a single dialog turn in a spoken dialog of the user input communication and the output at least one ranked sentence plan.

21. The tangible computer readable medium of claim 20, the instructions further comprising recognizing symbols in the user input communication, the symbols being used to generate the plurality of sentence plans based on the plurality of communicative goals.