Dialogue processing system and method

• US 7,020,607 B2
• Filed: 12/20/2000
• Issued: 03/28/2006
• Est. Priority Date: 07/13/2000
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Adachi, Akira
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US09/739,644
• Publication Number:
  US 20020038213A1
• Time in Patent Office: 1,924 Days
• Field of Search: 704/2, 704/5, 704/10, 704/257
• US Class Current: 704/257
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06Q 10/02 Reservations, e.g. for tickets, services or events
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

An object of the present invention is to realize a natural dialogue with a system having a simple configuration. The dialogue processing system of the present invention to perform a dialogue with a user, comprises: a plurality of slots (slot 11), wherein each of the slots is a storage area storing a preset information item (for example, boarding date, departure point, destination, or flight name) that are required to achieve the purpose of the dialogue with the user (for example, an airline ticket reservation); an analyzer (the speech details analyzer 19) for extracting information corresponding to the information items from information (for example, free sentence entered in natural language) entered by the user in the dialogue and for storing the extracted information in the slot for that information item; and a response processor (the dialogue processor 21) for outputting response information (a sentence prompting the input of insufficient information) to the user in accordance with information storage state of the slots.

Claims

1. A dialogue processing system for performing a natural language dialogue with a user, comprising:

a predetermined number of slots, each said slot being a storage area for storing one predefined information item which is required to perform a processing for a service requested by said user;

an analyzer extracting information corresponding to said information item from information entered by a certain user in said one dialogue and storing the extracted information in the slot for the information item;

a knowledge base storing information relating to said predefined information items, and

a response processor determining whether said one dialogue is completed based on whether all of said information items required to perform said processing for said service requested by said user have been stored into all of said predetermined number of slots, and outputting response information to said certain user in accordance with an information storage state of said slots

wherein said response processor comprises a checker that checks whether there is inconsistency between information stored in said slots and information stored in said knowledge base by using said information stored in said slots and said information stored in said knowledge base, and

wherein progress of said one dialogue is controlled based on said information storage state of said predetermined number of slots without using any dialogue sequence definition.

2. The dialogue processing system as set forth in claim 1, further comprising:

a knowledge base processor extracting information corresponding to information items required to perform said processing for said service requested by said user with said certain user by using information stored in said slots and information stored in said knowledge base and storing the extracted information in the slot for the required information item.

3. The dialogue processing system as set forth in claim 1, wherein said information entered by said certain user in said one dialogue is a sentence in natural language, and

wherein said analyzer comprises:

means for performing morpheme analysis for said sentence;

means for performing parsing processing for results of said morpheme analysis; and

means for extracting information corresponding to the information item based on the results of said morpheme analysis and said parsing processing by using an extraction rule pre-defined to perform said processing for said service requested by said user.

4. The dialogue processing system as set forth in claim 1, further comprising:

means for determining whether information stored in said knowledge base is necessary to be updated if all said information items required to perform said processing for said service requested by said user with said certain user are stored in all of said predetermined number of slots; and

means for updating said knowledge base in accordance with a predetermined rule if it is determined that said information stored in said knowledge base is necessary to be updated.

5. The dialogue processing system as set forth in claim 1, wherein said response processor comprises:

a response information storage device for storing response information for said user in correspondence with said information storage state of said slots; and

means for determining said information storage state of said slots and for acquiring and outputting response information for said certain user in correspondence with said information storage state of said slots from said response information storage device.

6. The dialogue processing system as set forth in claim 1, wherein said information entered by said certain user in said one dialogue is voice information entered in natural language, and

wherein said dialogue processing system further comprises:

a voice recognition processor for converting said voice information into character information; and

a voice synthesis processor for converting said response information into voice information.

7. The dialogue processing system as set forth in claim 1, wherein, in accordance with said information storage state of said slots, said response processor outputs response information for requesting said certain user to enter information items that are required to perform said processing for a service requested by said user.

8. A storage medium for storing a program for processing a natural language dialogue with a user, said program configured to perform a process, the process comprising:

ensuring a predetermined number of slots on a storage device, each said slot being a storage area for storing a predefined information item which is required to perform a processing for a service requested by said user;

extracting information corresponding to said information item from information entered by a certain user in said one dialogue and storing the extracted information in the slot for the information item;

a knowledge base storing information relating to said predefined information items, and

determining whether said one dialogue is completed based on whether all of said information items required to perform said processing for said service requested by said user have been stored into all of said predetermined number of slots, and outputting response information to said certain user in accordance with an information storage state of said slots, and

wherein said response processor comprises a checker that checks whether there is inconsistency between information stored in said slots and information stored in said knowledge base by using said information stored in said slots and said information stored in said knowledge base, and

wherein progress of said one dialogue is controlled based on said information storage state of said predetermined number of slots without using any dialogue sequence definition.

9. The storage medium as set forth in claim 8, said process further comprising:

extracting information corresponding to information items required to perform said processing for said service requested by said user with said certain user by using information stored in said slots and information stored in a knowledge base relating to said dialogue with said user; and

storing the extracted information in the slot for the required information item.

10. The storage medium as set forth in claim 8, said process further comprising:

checking whether there is inconsistency between information stored in said slots and information stored in a knowledge base relating to said dialogue with said user by using said information stored in said slots and said information stored in said knowledge base.

11. The storage medium as set forth in claim 8, wherein said information entered by said certain user in said one dialogue is a sentence in natural language, and wherein said extracting comprises:

performing morpheme analysis for said sentence;

performing parsing processing for results of said morpheme analysis; and

extracting information corresponding to the information item based on the results of said morpheme analysis and said parsing processing by using an extraction rule pre-defined to perform said processing for said service requested by said user.

12. The storage medium as set forth in claim 8, said process further comprising:

determining whether information stored in a knowledge base relating to said dialogue with said user is necessary to be updated if all said information items required to perform said processing for said service requested by said user with said certain user are stored in all of said predetermined number of slots; and

updating said knowledge base in accordance with a predetermined rule if it is determined that said information stored in said knowledge base is necessary to be updated.

13. The storage medium as set forth in claim 8, wherein said determining and outputting comprises:

determining said information storage state of said slots; and

acquiring and outputting response information for said certain user in correspondence with said information storage state of said slots from a response information storage device for storing response information for a user in correspondence with said information storage state of said slots.

14. The storage medium as set forth in claim 8,

wherein said information entered by said certain user in said one dialogue is voice information entered in natural language, and

wherein said process further comprises:

converting said the voice information into character information; and

converting said response information into voice information.

15. The storage medium as set forth in claim 8, wherein, in accordance with said information storage state of said slots, said outputting comprises outputting response information for requesting said certain user to enter information items that are required to perform said processing for a service requested by said user.

16. A method for processing a natural language dialogue with a user, said method comprising:

ensuring a predetermined number of slots on a storage device, each said slot being a storage area for storing a predefined information item which is required to perform a processing for a service requested by said user;

extracting information corresponding to said information item from information entered by a certain user in said one dialogue and storing the extracted information in the slot for the information item;

a knowledge base storing information relating to said predefined information items, and

determining whether said one dialogue is completed based on whether all of said information items required to perform said processing for a service requested by said user have been stored into all of said predetermined number of slots, and outputting response information to said certain user in accordance with an information storage state of said slots, and

wherein said response processor comprises a checker that checks whether there is inconsistency between information stored in said slots and information stored in said knowledge base by using said information stored in said slots and said information stored in said knowledge base, and

wherein progress of said one dialogue is controlled based on said information storage state of said predetermined number of slots without using any dialogue sequence definition.

17. The method as set forth in claim 16, further comprising:

extracting information corresponding to information items required to perform said processing for said service requested by said user with said certain user by using information stored in said slots and information stored in a knowledge base relating to said dialogue with said user; and

storing the extracted information in the slot for the required information item.

18. The method as set forth in claim 16, further comprising:

checking whether there is inconsistency between information stored in said slots and information stored in a knowledge base relating to said dialogue with said user by using said information stored in said slots and said information stored in said knowledge base.

19. The method as set forth in claim 16, wherein said information entered by said certain user in said one dialogue is a sentence in natural language, and

wherein said extracting comprises:

performing morpheme analysis for said sentence;

performing parsing processing for results of said morpheme analysis; and

extracting information corresponding to the information item based on the results of said morpheme analysis and said parsing processing by using an extraction rule pre-defined to perform said processing for said service requested by said user.

20. The method as set forth in claim 16, further comprising:

determining whether information stored in a knowledge base relating to said dialogue with said user is necessary to be updated if all said information items required to perform said processing for said service requested by said user with said certain user are stored in all of said predetermined number of slots; and

updating said knowledge base in accordance with a predetermined rule if it is determined that said information stored in said knowledge base is necessary to be updated.

21. The method as set forth in claim 16, wherein said determining and outputting comprises:

determining said information storage state of said slots; and

acquiring and outputting response information for said certain user in correspondence with said information storage state of said slots from a response information storage device for storing response information for a user in correspondence with said information storage state of said slots.

22. The method as set forth in claim 16, wherein said information entered by said certain user in said one dialogue is voice information entered in natural language, and wherein said method further comprises:

converting said the voice information into character information; and

converting said response information into voice information.

23. The method as set forth in claim 16, wherein, in accordance with said information storage state of said slots, said determining and outputting comprises outputting response information for requesting said certain user to enter information items that are required to perform said processing for a service requested by said user.

24. A method of processing a natural language dialogue with a user, the method comprising:

providing a predetermined dialog objective, the predetermined dialog objective comprising a set slots of information, each slot comprising a specific field of information that needs to be filled to complete the dialog objective;

preparing or selecting natural language prompts to be presented to the user, where the prompts are prepared or selected based on which slots have filled fields and which do not;

storing information relating to said predefined information items in a knowledge base, and

receiving sentence-like natural language responses to the natural language prompts and, for each response, using natural language processing to automatically extract one or more fields of information that the user intended to convey with the response, where, for at least one received response, two different fields are extracted and determined to correspond to slot fields, where at least one of the extracted fields was supplemental information volunteered from the user without having been previously specifically prompted for by a prompt prepared or selected to be presented to the user, whereby progress of the natural language dialogue is controlled based on the information storage state of the predetermined number of slots without reference to any dialogue sequence definition.

25. A method according to claim 24, further comprising:

allowing a dynamic order of presenting the prompts to the user.

26. A method according to claim 24, further comprising:

given a slot previously mentioned during the dialog, allowing the user or a system preparing the prompts to use a pronoun to refer to the previously mentioned slot, where, for some later response having the pronoun, the system automatically determines which slot is referred to by the pronoun in the later response.

27. A method according to claim 24, further comprising:

automatically determining that the dialog objective has been met when all of the slot fields are filled.