JP6849964B2 - Dialogue control device, dialogue engine, management terminal, dialogue device, dialogue control method, dialogue method, and program - Google Patents

Description

The present invention relates to a dialogue control device or the like that controls dialogue.

Conventionally, there has been a voice dialogue control device for the purpose of improving the flexibility of communication with a user. The control of the voice dialogue device corrects the generated response when it is determined that the correction condition for determining whether or not the response needs to be corrected is satisfied in the waiting time from the acquisition of the voice to the generation of the response. It is provided with a correction execution unit (see Patent Document 1).

Further, conventionally, there has been a highly maintainable dialogue system in which a classification category suitable for an actual user's utterance can be set relatively easily (see Patent Document 2). Such a dialogue system advances the dialogue by using the state transition information of the dialogue.

Japanese Unexamined Patent Publication No. 2015-2569 Japanese Unexamined Patent Publication No. 2011-215742

However, in the conventional device, since only one dialogue engine is used, there are cases where an appropriate dialogue cannot be continued.

Also, for example, even though the dialogue between the user and the dialogue engine is broken, there is no choice but to continue the dialogue or end the dialogue.

The dialogue control device of the first invention is a dialogue control device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines, and obtains user utterance information corresponding to the user's utterance. A receiving unit received from the user terminal, an utterance information output unit capable of outputting utterance information corresponding to the user utterance information to one or more dialogue engines, and one or more utterance information depending on the output of the utterance information. A response that transmits to the user terminal an automatic response reception unit that receives automatic response information from each dialogue engine and one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit. It is a dialogue control device including a transmission unit.

With such a configuration, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

Further, in the dialogue control device of the second invention, the utterance information output unit outputs the utterance information to two or more dialogue engines, and the automatic response reception unit outputs the utterance information with respect to the first invention. The response transmission unit further includes a selection unit that receives automatic response information from two or more dialogue engines and selects one automatic response information from two or more automatic response information, and the response transmission unit is an automatic unit selected by the selection unit. It is an interactive control device that transmits one automatic information corresponding to the response information to the user terminal.

With such a configuration, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

Further, in the dialogue control device of the third invention, with respect to the second invention, the selection unit selects one automatic response information that matches a predetermined condition from two or more automatic response information. It is a control device.

With such a configuration, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

Further, in the dialogue control device of the fourth invention, with respect to the second or third invention, the selection unit is among the automatic response information satisfying the predetermined conditions among the two or more automatic response information. , A dialogue control device that selects the automatic response information output from the dialogue engine with the highest priority based on the priority information.

With such a configuration, one response can be appropriately selected from a plurality of dialogue engines, and an appropriate dialogue can be continued.

Further, the dialogue control device of the fifth invention further includes a selection unit for selecting one dialogue engine from two or more dialogue engines with respect to the first invention, and the utterance information output unit outputs utterance information. , The selection unit outputs to one dialogue engine selected, the automatic response reception unit receives the automatic response information from one dialogue engine according to the output of the utterance information, and the response transmission unit receives the automatic response reception unit. It is a dialogue control device that transmits one automatic information corresponding to the automatic response information to the user terminal.

With such a configuration, one dialogue engine can be appropriately selected from a plurality of dialogue engines, and an appropriate dialogue can be continued.

Further, the dialogue control device of the sixth invention is a dialogue control device for selecting one dialogue engine by using the user utterance information or the utterance information with respect to the fifth invention.

With such a configuration, one dialogue engine can be appropriately selected from a plurality of dialogue engines by using the user's utterance, and an appropriate dialogue can be continued.

Further, the dialogue control device of the seventh invention can output the utterance information to the management terminal for any one of the first to sixth inventions, and the utterance information output unit can output the utterance information to the management terminal. It further includes a manual response receiving unit that receives manual response information that is a response from the management terminal in response to output to the management terminal, and the response transmitting unit is one automatic information that is information corresponding to one automatic response information. Alternatively, it is an interactive control device that transmits manual information, which is information corresponding to manual response information, to a user terminal.

With such a configuration, an appropriate dialogue can be continued by properly using the response from the dialogue engine and the response from the administrator and outputting them to the user terminal.

Further, the dialogue control device of the eighth invention further includes a decision unit for deciding whether to adopt a dialogue engine or a management terminal, and the response transmission unit determines that the decision adopts the dialogue engine. In the case of, the interactive control device according to claim 7, wherein one automatic information is transmitted to the user terminal, and in the case of the decision to adopt the management terminal, the manual information is transmitted to the user terminal.

With such a configuration, an appropriate dialogue can be continued by properly using the response from the dialogue engine and the response from the administrator and outputting them to the user terminal.

Further, the dialogue control device of the ninth invention stores mode information that can at least adopt either an automatic mode that employs a dialogue engine or a manual mode that employs a management terminal for the eighth invention. Mode storage unit, a judgment unit that determines whether or not a predetermined condition for changing the mode information is satisfied, and a judgment unit that determines whether or not a predetermined condition is satisfied, the mode information is changed. The mode changing unit is further provided, and the determining unit is an interactive control device that acquires the mode information of the mode storage unit.

With such a configuration, an appropriate dialogue can be continued by properly using the response from the dialogue engine and the response from the administrator and outputting them to the user terminal.

Further, the dialogue control device of the tenth invention stores mode information that can at least take either an automatic mode adopting a dialogue engine or a manual mode adopting a management terminal for the eighth invention. It further includes a mode storage unit, a change instruction reception unit that receives a change instruction that is a mode change instruction from a management terminal, and a mode change unit that changes the mode information of the mode storage unit in response to the change instruction. The determination unit is a dialogue control device that determines whether to adopt the dialogue engine or the management terminal according to the mode information of the mode storage unit.

With such a configuration, an appropriate dialogue can be continued by properly using the response from the dialogue engine and the response from the administrator and outputting them to the user terminal.

Further, the dialogue control device of the eleventh invention is a dialogue control device further comprising an automatic information output unit for outputting one automatic information to a management terminal for any one of the seventh to tenth inventions. is there.

With such a configuration, an appropriate dialogue can be continued by properly using the response from the dialogue engine and the response from the administrator and outputting them to the user terminal.

Further, the dialogue control device of the twelfth invention is a dialogue control device further provided with a manual information output unit for outputting manual information to the dialogue engine for any one of the seventh to eleventh inventions. ..

With such a configuration, it is possible to contribute to the improvement of the capacity of the dialogue engine by passing the response from the administrator to the dialogue engine.

Further, the dialogue engine of the thirteenth invention stores learning information which is information for acquiring automatic response information corresponding to speech information for any one of the first to twelfth inventions. A learning information storage unit, a utterance reception unit that receives utterance information from the dialogue control device, a response acquisition unit that acquires automatic response information that is a response corresponding to the utterance information received by the utterance reception unit, and a response acquisition unit that uses the learning information. A response output unit that outputs the automatic response information acquired by the response acquisition unit to the dialogue control device, a learning reception unit that receives other automatic response information that is a response from another dialogue engine, and another automatic response received by the learning reception unit. It is a dialogue engine including a learning unit that adds information to learning information using information.

With such a configuration, it is possible to provide a dialogue engine capable of learning the response from another dialogue engine.

Further, the dialogue device of the fourteenth invention is a dialogue device capable of receiving automatic response information which is a response corresponding to a user's utterance from two or more dialogue engines, and is information corresponding to the user's utterance information. An utterance information output unit that can output utterance information to one or more dialogue engines, an automatic response reception unit that receives automatic response information from one or more dialogue engines according to the output of utterance information, and an automatic response reception unit. It is an interactive device including a response output unit that transmits one automatic information corresponding to one of the one or more automatic response information received by the unit to the user terminal.

With such a configuration, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

Further, in the dialogue device of the fifteenth invention, with respect to the fourteenth invention, the utterance information output unit outputs the utterance information regarding the user utterance information to at least one of one or more dialogue engines or screens. The response output unit is further provided with a manual response reception unit that outputs the utterance information to the screen and receives the manual response information that is the response from the administrator, and the response output unit is the automatic information or the manual about the automatic response information. It is an interactive device that outputs manual information about response information.

With such a configuration, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

According to the dialogue control device according to the present invention, a plurality of dialogue engines can be effectively used to continue an appropriate dialogue.

Conceptual diagram of the dialogue system A in the first embodiment Block diagram of the dialogue system A Block diagram of the dialogue control device 2 A flowchart for explaining the operation of the dialogue control device 2. Flow chart explaining the operation of the dialogue engine 3 Diagram showing the same priority information management table Diagram showing the learning information management table Diagram showing the same Foodtable The figure which shows the output example of the user terminal 1. Block diagram of the dialogue control device 4 according to the second embodiment A flowchart for explaining the operation of the dialogue control device 4. The figure which shows the selection condition management table Conceptual diagram of the dialogue system C in the third embodiment Block diagram of the dialogue system C Block diagram of the dialogue control device 5 A flowchart for explaining the operation of the dialogue control device 5. A flowchart explaining the operation of the dialogue engine 6. Flow chart explaining the operation of the management terminal 7. Diagram showing the bankruptcy judgment condition management table Diagram showing the same output example The figure which shows the manual response information management table Block diagram of the dialogue device 8 according to the fourth embodiment A flowchart illustrating the operation of the dialogue device 8. Overview of the computer system according to the above embodiment Block diagram of the computer system

Hereinafter, embodiments of the dialogue control device and the like will be described with reference to the drawings. In addition, since the components with the same reference numerals perform the same operation in the embodiment, the description may be omitted again.

(Embodiment 1)
In the present embodiment, a dialogue system that sends a request to a plurality of dialogue engines and adopts one response from the plurality of responses will be described.

In the present embodiment, a dialogue system that employs the response of the dialogue engine that meets the predetermined conditions will be described.

In the present embodiment, a dialogue system that adopts the response of the dialogue engine having the highest priority to the dialogue engine among the responses satisfying the predetermined conditions will be described.

The dialogue control device constituting the dialogue system in the present embodiment is a device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines.

FIG. 1 is a conceptual diagram of the dialogue system A according to the present embodiment. The dialogue system A includes one or more user terminals 1, a dialogue control device 2, and two or more dialogue engines 3. The user terminal 1 is, for example, a so-called personal computer, a tablet terminal, a smartphone, or the like, and the type thereof does not matter. The user terminal 1 is a terminal used by the user. A user is a person who interacts with a dialogue engine. The dialogue control device 2 is a device that controls dialogue. The dialogue control device 2 is, for example, an ASP server, a cloud server, or the like. However, the type of the dialogue control device 2 does not matter. Each of the two or more dialogue engines 3 is a device that receives information on utterances from the user and returns a response. The dialogue engine 3 is, for example, an ASP server, a cloud server, or the like. However, the type of the dialogue engine 3 does not matter. Further, the dialogue engine 3 receives the utterance information and outputs the automatic response information. The dialogue engine 3 may be a known dialogue engine.

Further, usually, one or more user terminals 1 and the dialogue control device 2 can communicate with each other via a network such as the Internet or LAN. Further, usually, the dialogue control device 2 and each of the dialogue engines 3 of 1 or more can communicate with each other via a network such as the Internet or LAN. However, the dialogue control device 2 may include one or more devices of one or more user terminals 1 and one or more dialogue engines 3 inside. That is, two or more of the user terminal 1, the dialogue control device 2, and the dialogue engine 3 may be physically realized by one device.

FIG. 2 is a block diagram of the dialogue system A according to the present embodiment. FIG. 3 is a block diagram of the dialogue control device 2.

The user terminal 1 includes a user reception unit 11, a user processing unit 12, a user transmission unit 13, a user reception unit 14, and a user output unit 15.

The dialogue control device 2 includes a storage unit 21, a reception unit 22, a processing unit 23, an output unit 24, a reception unit 25, and a transmission unit 26. The processing unit 23 includes a selection unit 231. The output unit 24 includes an utterance information output unit 241. The reception unit 25 includes an automatic response reception unit 251. The transmission unit 26 includes a response transmission unit 261.

The dialogue engine 3 includes a learning information storage unit 31, an utterance reception unit 32, a response acquisition unit 33, a response output unit 34, a learning reception unit 35, and a learning unit 36.

The user reception unit 11 constituting the user terminal 1 receives an utterance from the user. The utterance may be accepted by voice or by text. The utterance may be information uttered by the user or information input by the user. The utterance input means may be any, such as a microphone, a keyboard, a mouse, and a menu screen. The user reception unit 11 can be realized by a device driver for input means such as a microphone or a keyboard, control software for a menu screen, or the like.

The user processing unit 12 acquires the user utterance information corresponding to the utterance from the utterance received by the user reception unit 11. For example, the user processing unit 12 voice-recognizes utterances that are voices and acquires user utterance information that is text. The user utterance information may be the same as the utterance input by the user.

Further, when the automatic information received by the user receiving unit 14 is text, the user processing unit 12 may generate voice data from the automatic information.

The user transmission unit 13 transmits the user utterance information acquired by the user processing unit 12 to the dialogue control device 2.

The user receiving unit 14 receives the automatic information described later. The automatic information is usually transmitted from the dialogue control device 2.

The user output unit 15 outputs the automatic information received by the user reception unit 14. The user output unit 15 may output the voice data acquired from the automatic information by the user processing unit 12. Even in such a case, it is assumed that the automatic information is output. Further, the output here is usually a display on a display or an audio output to a speaker, but projection using a projector, printing by a printer, transmission to an external device, storage on a recording medium, It may be considered that the concept includes delivery of the processing result to another processing device or another program.

Various types of information are stored in the storage unit 21 that constitutes the dialogue control device 2. The various types of information include, for example, selection conditions for automatic response information described later, priority information described later, dialogue history, and information for communicating with each of two or more dialogue engines 3 (for example, the IP address of the dialogue engine 3). , URL, MAC address, etc.). The dialogue history is, for example, user utterance information, utterance information, automatic response information, automatic information, and the like.

The receiving unit 22 receives the user utterance information from the user terminal 1. Further, the user utterance information is usually text, but voice data may also be used. The data structure and data type of user utterance information do not matter.

The processing unit 23 performs various processes. The various processes are processes performed by the selection unit 231 and the like.

The selection unit 231 selects one automatic response information from two or more automatic response information. Such a selection method may be random. Random is to arbitrarily select one automatic response information from two or more automatic response information.

Further, the selection unit 231 selects one automatic response information that matches a predetermined condition from two or more automatic response information. The predetermined conditions are, for example, a score-based condition, a content-based condition of the automatic response information, a time-based condition, a priority-based condition, and the like, or a combination thereof. The condition based on the score is, for example, the condition that the score corresponding to the automatic response information is the highest. Further, the condition based on the content of the automatic response information is, for example, the automatic response information of the longest character string or the automatic response information having the largest number of sentences. The time-based condition is, for example, the condition that the automatic response information is received earliest. Further, the priority-based condition is, for example, the automatic response information output from the dialogue engine 3 having the highest priority based on the priority information among the automatic response information received within the threshold time. It is a condition. The conditions are not limited. When based on the priority, the storage unit 21 usually stores the priority information. The priority information is information indicating the priority of two or more dialogue engines 3. The priority information is, for example, an identifier of the dialogue engine 3 arranged in descending order of priority. The priority information is, for example, information having two or more sets of a priority (for example, a numerical value) and an identifier of the dialogue engine 3.

The selection unit 231 selects the automatic response information output from the dialogue engine 3 having the highest priority based on the priority information among the automatic response information satisfying the predetermined conditions among the two or more automatic response information. To do. Here, the automatic response information satisfying the predetermined conditions is, for example, automatic response information having no information indicating the failure of the dialogue (for example, "incomprehensible" or "incomprehensible"), and automatic response with a score equal to or higher than the threshold value. Response information, etc.

The algorithm (predetermined condition described above) in which the selection unit 231 selects one automatic response information from two or more automatic response information does not matter.

The output unit 24 outputs various information. The output here is usually a transmission. However, the output means that the information should be passed to the output destination.

The utterance information output unit 241 outputs utterance information, which is information corresponding to the user utterance information, to each of one or more dialogue engines 3. Here, a case where the utterance information output unit 241 mainly outputs the utterance information to each of two or more dialogue engines 3 will be described. Here, the utterance information may be the same information as the user utterance information. Further, the utterance information may be a text obtained by voice-recognizing the user utterance information. It is preferable that the utterance information is information to which the utterance identifier is added. The utterance identifier is information that identifies the utterance.

The automatic response information output unit 242 outputs one automatic response information selected by the selection unit 231 to one or more dialogue engines 3 other than the dialogue engine 3 that outputs the one automatic response information. The automatic response information output unit 242 may output one or more automatic response information received by the automatic response reception unit 251 to all other dialogue engines 3 except for the automatic response information output by itself. However, it may be output to some of the dialogue engines 3 among the other dialogue engines 3. Further, the automatic response information output unit 242 outputs the automatic response information to the dialogue engine 3 for learning of the dialogue engine 3.

The reception unit 25 receives various information, instructions, and the like. Acceptance here is usually reception. However, the reception may be as long as the information etc. is received.

The automatic response reception unit 251 receives one or two or more automatic response information which is a response from the dialogue engine 3 according to the output of the utterance information to the dialogue engine 3. Normally, the automatic response reception unit 251 receives automatic response information from two or more dialogue engines according to the output of the utterance information. The reception here is, for example, reception.

The transmission unit 26 transmits various types of information. The various types of information are, for example, automatic information and the like.

The response transmission unit 261 transmits one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit 251 to the user terminal 1.

The response transmission unit 261 normally transmits one automatic information corresponding to the automatic response information selected by the selection unit 231 to the user terminal 1.

The learning information storage unit 31 constituting the dialogue engine 3 stores one or more learning information. When the term "learning information" is simply used, two or more learning information may be used. Further, the learning information is information for acquiring the automatic response information corresponding to the utterance information. The learning information is, for example, a set of utterance information and automatic response information. Further, the learning information is, for example, a set of a character string pattern of utterance information and a character string pattern of automatic response information. Further, the learning information may be, for example, information indicating a state transition of conversation (see, for example, Patent Document 2). Further, the learning information may include a dictionary of words and the like. Needless to say, the data structure and contents of the learning information do not matter.

The utterance reception unit 32 receives utterance information from the dialogue control device 2. The utterance reception unit 32 usually receives utterance information from the dialogue control device 2.

The response acquisition unit 33 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, by using the learning information of the learning information storage unit 31. The processing of the response acquisition unit 33 is a known technique. For example, the response acquisition unit 33 searches the learning information storage unit 31 for automatic response information paired with the utterance information received by the utterance reception unit 32. Further, for example, the response acquisition unit 33 detects a character string pattern of the utterance information that matches the utterance information received by the utterance reception unit 32, and the character string pattern of the automatic response information paired with the character string pattern of the utterance information. Is acquired, and the word or phrase extracted from the utterance information is substituted into the variable part of the character string pattern of the automatic response information to construct the automatic response information.

The response output unit 34 outputs the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2. The output here is usually a transmission. However, as for the output, the automatic response information may be passed to the dialogue control device 2.

The learning reception unit 35 receives automatic response information which is a response from another dialogue engine 3. The acceptance of the automatic response information here is intended to accept the automatic information.

The learning unit 36 performs learning processing using the automatic response information received by the learning reception unit 35. The learning process is, for example, adding information to the learning information of the learning information storage unit 31 by using the automatic response information. The learning process is, for example, changing the learning information of the learning information storage unit 31 by using the automatic response information. The process of adding information to the learning information includes a process of adding new learning information to the learning information storage unit 31.

The learning unit 36 adds, for example, learning information, which is a set of automatic response information and utterance information corresponding to the automatic response information, to the learning information storage unit 31.

For example, the learning unit 36 converts the information of the set of the automatic response information from the other dialogue engine 3 and the utterance information corresponding to the automatic response information into the learning information which is the state transition information, and automatically responds before the utterance information. A utterance information node is added as a state ahead of the information node (state), and an automatic response information node is added to the utterance information node.

For example, the learning unit 36 constitutes an automatic response information pattern from the automatic response information from the other dialogue engine 3, and constitutes an utterance information pattern from the utterance information corresponding to the automatic response information, and forms an utterance information pattern. The learning information, which is a set of the response pattern and the response pattern, is added to the learning information storage unit 31.

For example, the learning unit 36 configures a pattern of automatic response information from automatic response information from another dialogue engine 3, and configures a pattern of utterance information from utterance information corresponding to the automatic response information, and sets the pattern. The learning information of the learning information storage unit 31 is updated so as to increase the appearance frequency or appearance probability of the pair of the utterance information pattern and the automatic response information pattern that matches. In such a case, the learning information has a set of a pattern of utterance information and a pattern of automatic response information, and an appearance frequency or appearance probability.

The learning unit 36 performs the above learning process using, for example, the utterance information and the automatic response information in which the utterance identifier corresponding to the utterance information and the utterance identifier corresponding to the automatic response information are the same.

In addition, there may be various learning processes using the automatic response information from the other dialogue engine 3. The learning process performed by the learning unit 36 may use another known learning processing algorithm, except that the automatic response information from the other dialogue engine 3 is used.

The user processing unit 12, the processing unit 23, the selection unit 231 and the response acquisition unit 33, and the learning unit 36 can usually be realized from an MPU, a memory, or the like. The processing procedure of the user processing unit 12 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

The user transmission unit 13, the output unit 24, the utterance information output unit 241, the transmission unit 26, the response transmission unit 261 and the response output unit 34 are usually realized by wireless or wired communication means, but are realized by broadcasting means. You may.

The output unit 24, the utterance information output unit 241, the automatic information output unit 242, and the response output unit 34 may be realized by output means for outputting information. The output of information here is, for example, the delivery of information by a program.

The user receiving unit 14, the receiving unit 22, the receiving unit 25, the automatic response receiving unit 251, the utterance receiving unit 32, and the learning receiving unit 35 are usually realized by wireless or wired communication means, but are means for receiving broadcasts. It may be realized by.

The reception unit 25, the automatic response reception unit 251 and the utterance reception unit 32, and the learning reception unit 35 may be realized by the reception means for receiving information. The acceptance of information here is, for example, the delivery of information by a program.

The user output unit 15 may or may not include an output device such as a display or a speaker. The user output unit 15 can be realized by the driver software of the output device, the driver software of the output device, the output device, or the like.

The storage unit 21 and the learning information storage unit 31 are preferably non-volatile recording media, but can also be realized by volatile recording media.

The process of storing information in the storage unit 21 or the like does not matter. For example, the information may be stored in the storage unit 21 or the like via the recording medium, or the information transmitted via the communication line or the like may be stored in the storage unit 21 or the like. Alternatively, the information input via the input device may be stored in the storage unit 21 or the like.

Next, the operation of the dialogue system A will be described. First, an operation example of the user terminal 1 will be described.

The user reception unit 11 of the user terminal 1 receives an utterance from the user. Next, the user processing unit 12 acquires the user utterance information corresponding to the utterance from the utterance received by the user reception unit 11. Here, the user processing unit 12 recognizes a voice utterance, for example, and acquires the text user utterance information. Next, the user transmission unit 13 transmits the user utterance information acquired by the user processing unit 12 to the dialogue control device 2. Next, in response to the transmission of the user utterance information, the user receiving unit 14 receives the automatic information. The automatic information is information on the response to the user's utterance. Next, the user output unit 15 outputs the automatic information received by the user reception unit 14. By repeating the above process, the dialogue between the user and the dialogue engine 3 is continued. When the automatic information is text, the user output unit 15 may voice-synthesize the text and output it as voice. In addition, the dialogue ends according to the user's instruction or the like.

Next, the operation of the dialogue control device 2 will be described with reference to the flowchart of FIG.

(Step S401) The receiving unit 22 determines whether or not the user utterance information has been received from the user terminal 1. If the user utterance information is received, the process proceeds to step S402, and if not received, the process returns to step S401.

(Step S402) The processing unit 23 acquires the utterance information from the user utterance information.

(Step S403) The output unit 24 acquires two or more output destinations of the utterance information. Such an output destination is two or more dialogue engines 3, and is stored in, for example, a storage unit 21. The output destination is, for example, a communication destination identifier described later.

(Step S404) The utterance information output unit 241 outputs the utterance information acquired in step S402 to two or more output destinations acquired in step S403.

(Step S405) The reception unit 25 receives one or more automatic response information in response to the output of the utterance information to the two or more dialogue engines 3. It is preferable that the reception unit 25 receives the automatic response information only for a time within the threshold value from the output of the utterance information.

(Step S406) The selection unit 231 selects one automatic response information from one or more automatic response information received in step S405. If there is only one automatic response information received in step S405, this step may be omitted.

(Step S407) The processing unit 23 acquires automatic information from the automatic response information selected in step S406.

(Step S408) The response transmission unit 261 transmits the automatic information acquired in step S407 to the user terminal 1.

(Step S409) The automatic response information output unit 242 outputs the one automatic response information selected in step S406 to the dialogue engine 3 other than the dialogue engine 3 that outputs the one automatic response information. Return to step S401.

In the flowchart of FIG. 4, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Next, the operation of the dialogue engine 3 will be described with reference to the flowchart of FIG.

(Step S501) The utterance reception unit 32 determines whether or not the utterance information has been received from the dialogue control device 2. If the utterance information is accepted, the process goes to step S502, and if the utterance information is not accepted, the process goes to step S504.

(Step S502) The response acquisition unit 33 acquires the automatic response information corresponding to the utterance information received in step S501 by using the learning information of the learning information storage unit 31.

(Step S503) The response output unit 34 outputs the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2.

(Step S504) The learning reception unit 35 determines whether or not the automatic response information of the learning target has been received. If the automatic response information is accepted, the process proceeds to step S505, and if the automatic response information is not accepted, the process returns to step S501.

(Step S505) The learning unit 36 configures information to be added to the learning information by using the automatic response information received by the learning reception unit 35.

(Step S506) The learning unit 36 adds the information configured in step S505 to the learning information storage unit 31. Return to step S501.

In the flowchart of FIG. 5, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Hereinafter, the specific operation of the dialogue system A in the present embodiment will be described. In this specific example, it is assumed that the dialogue system A includes three dialogue engines. The conceptual diagram of the dialogue system A is FIG.

Now, the storage unit 21 of the dialogue control device 2 stores the priority information management table shown in FIG. The priority information management table is a table that manages the priority and the like of each dialogue engine 3. The priority information management table has three records having "priority", "interaction engine identifier", and "notification destination identifier". The smaller the "priority" is, the higher the priority is. The "notification destination identifier" is an IP address here, but may be any information as long as it is information for the dialogue control device 2 to communicate with the dialogue engine 3.

Further, learning that constitutes one of the three dialogue engines 3 (for example, the dialogue engine 3 identified by "e1"; hereinafter, such a dialogue engine is referred to as the dialogue engine "e1"). It is assumed that the information storage unit 31 stores the learning information management table shown in FIG. 7. The learning information management table manages one or more records having "ID" and "learning information". The "ID" is information that identifies a record. "Learning information" is a pair of "utterance information" and "automatic response information". The "utterance information" and "automatic response information" may be constants (fixed character strings) or information having variables. Further, the "automatic response information" may be executable information such as a program or a search expression (here, for example, it is an SQL statement, but other formats may be used). {} In the learning information indicates a variable, and * in the learning information indicates 0 or more characters. Further, $ in the automatic response information is a variable, and the value is held in the learning information storage unit 31. The value of "$ business day" is assumed to be "Sunday, Tuesday, Wednesday, Thursday, Friday, Saturday" here. Further, it is assumed that "include" in the automatic response information is a logical operator of whether or not it is included. That is, "{timeExpress} include $ business day" indicates whether or not the information of the variable "timeExpression" is included in the variable "business day". Then, if it is included, a true "1" is returned, and if it is not included, a false "0" is returned.

Further, it is assumed that the learning information storage unit 31 of the dialogue engine "e1" stores the Foodtable shown in FIG. Foodtable stores one or more records having "ID", "cooking", and "how to make".

Further, it is assumed that the learning information storage unit 31 of the dialogue engine "e1" has a food dictionary for storing the dish name and a crops dictionary for storing the crop name. The food dictionary stores a large number of dish names such as "Katsudon, Gyudon, Oyakodon, ...". In addition, a large number of crop names such as "rice, eggplant, carrot, ..." are stored in the crops dictionary.

In such a situation, the following three specific examples will be described. Specific example 1 is a case where the automatic response information is selected based on the priority. More specifically, in the specific example 1, the dialogue control device 2 outputs an automatic response output from the dialogue engine 3 having the highest priority based on the priority information among the automatic response information satisfying the predetermined conditions. When selecting information. Further, the predetermined condition here is that, for example, the automatic response information, which is the response from the dialogue engine 3, does not include predetermined terms (for example, "did not understand" or "incomprehensible"). is there.

Specific example 2 is a case based on a time condition. It may be said that the time condition is based on the speed at which the automatic response information is received. More specifically, the second embodiment is a case where the dialogue control device 2 selects the first received automatic response information that satisfies a predetermined condition. The predetermined conditions here also assume that, for example, the automatic response information that is the response from the dialogue engine 3 does not include predetermined terms (for example, "did not understand" or "incomprehensible").

Specific example 3 is a case based on the content condition. The content condition is a condition based on the content of the automatic response information (response). More specifically, in the specific example 3, the dialogue control device 2 has the longest sentence length or the large number of independent words among the automatic response information satisfying the predetermined conditions. This is the case when the automatic response information with the largest number of sentences is selected. The predetermined conditions here also assume that, for example, the automatic response information that is the response from the dialogue engine 3 does not include predetermined terms (for example, "did not understand" or "incomprehensible").

(Specific example 1)
Now, it is assumed that the user voice-inputs "Tell me how to make beef bowl" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Tell me how to make beef bowl" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user's utterance information "Tell me how to make beef bowl". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information "Tell me how to make beef bowl" from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "1: Tell me how to make beef bowl" from the user's utterance information. The "1" in the utterance information is the utterance identifier.

Next, the utterance information output unit 241 acquires the output destinations (three communication destination identifiers) of the utterance information from the priority information management table of FIG. Then, the utterance information output unit 241 transmits the utterance information "1: Tell me how to make beef bowl" to the three dialogue engines 3 identified by each of the three communication destination identifiers.

Next, the utterance receiving unit 32 of each of the three dialogue engines 3 receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 of each of the three dialogue engines 3 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, using the learning information of the learning information storage unit 31. For example, the response acquisition unit 33 of the dialogue engine 3 identified by the dialogue engine identifier "e1" morphologically analyzes "tell me how to make beef bowl" contained in the utterance information. Then, the response acquisition unit 33 searches the dictionary using the acquired morpheme "gyudon" as a key, and detects that "gyudon" is included in {food}. Next, the response acquisition unit 33 detects that "Tell me how to make beef bowl" matches the utterance information of "ID = 4" in the learning information management table of FIG. 7. Then, the response acquisition unit 33 acquires the automatic response information of "ID = 4", executes the automatic response information, and prepares "Gyudon" corresponding to the dish "Gyudon" from the Foodtable of FIG. Introducing ... "is acquired. Then, the response output unit 34 transmits the automatic response information "1: 4 servings of material ...." acquired by the response acquisition unit 33 to the dialogue control device 2.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from the dialogue engine 3 received within the threshold time. Here, it is assumed that the reception unit 25 has received the automatic response information from the three dialogue engines 3 within the threshold time.

Next, the selection unit 231 selects one automatic response information from the three automatic response information. That is, the selection unit 231 does not select the automatic response information transmitted from the dialogue engine 3 of "e2" that includes a predetermined term ("unintelligible" or "incomprehensible") in the automatic response information. decide. Next, the priority information "1" and "3" of the dialogue engine 3 of "e1" and "e3" are acquired, and the automatic response of the dialogue engine 3 of "e1" having a high priority (priority information "1") is obtained. Select the information "1: 4 servings of ingredients ...".

Next, the processing unit 23 acquires the automatic information "Introducing the material for 4 servings ..." from the selected automatic response information.

Next, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information. Then, the user output unit 15 outputs the automatic information.

In addition, the automatic response information output unit 242 displays the selected automatic response information "1: 4 servings of material ...." other than the dialogue engine 3 that outputs the one automatic response information. Is transmitted to the dialogue engine 3 (dialogue engine 3 of "e2" and "e3").

Then, the learning reception unit 35 of the dialogue engine 3 of "e2" and "e3" receives the automatic response information.

Next, the learning unit 36 of the dialogue engine 3 of the “e2” and “e3” configures information to be added to the learning information by using the automatic response information received by the learning reception unit 35. That is, the learning unit 36 acquires the utterance information corresponding to the utterance identifier "1" possessed by the automatic response information "I will introduce the material for 1: 4 servings ....". Then, a set of the content of the utterance information corresponding to the utterance identifier "1" "Tell me how to make beef bowl" and the content of the self-response information "Introducing the ingredients for 4 servings ..." is shown in the figure. Add to the learning information management table of 8.

The learning unit 36 replaces the nouns or noun phrases "beef bowl" and "how to make" with variables in the content of the utterance information "tell me how to make beef bowl", and replaces "{variable 2} of {variable 1}" with variables. Change to the pattern "Tell me" and replace the noun or noun phrase "4 servings" and "material" with variables in "Introducing materials for 4 servings ...." You may change the pattern to "Introducing variable 4} ...." and add a set of two patterns to the learning information management table in Fig. 8.

That is, in this specific example, the learning information may be improved by using the automatic response information input from the other dialogue engine 3, and the structure of the learning information does not matter.

Next, it is assumed that the user voice-inputs "Which soup is recommended for powdered udon soup?" To the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the voice input of the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user utterance information "Which soup is recommended for powdered udon soup?". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "2: Which soup is recommended for the powdered udon soup?" From the user's utterance information.

Next, the utterance information output unit 241 acquires three communication destination identifiers from the priority information management table of FIG. Then, the utterance information output unit 241 transmits the utterance information to the three dialogue engines 3 identified by the three communication destination identifiers.

Next, the utterance receiving unit 32 of each of the three dialogue engines 3 receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 of each of the three dialogue engines 3 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, using the learning information of the learning information storage unit 31. Then, the response output unit 34 of each of the three dialogue engines 3 transmits the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from the dialogue engine 3 received within the threshold time.

Next, the selection unit 231 selects one automatic response information from the received automatic response information by the algorithm using the priority information described above. Then, the processing unit 23 acquires the automatic information from the one automatic response information.

Next, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information "Introducing the material for four servings ....". Then, the user output unit 15 outputs the automatic information.

Further, the automatic response information output unit 242 transmits the selected automatic response information to the dialogue engine 3 other than the dialogue engine 3 that outputs the one automatic response information.

Then, the learning reception unit 35 of the dialogue engine 3 that has not transmitted the one automatic response information receives the one automatic response information. The learning unit 36 performs the above-mentioned learning process using the automatic response information received by the learning reception unit 35.

By repeating the above processing, an appropriate dialogue can be continued. In addition, the user can obtain an appropriate response. Further, the dialogue engine 3 can learn the abilities (which may be called knowledge) of other dialogue engines 3.

(Specific example 2)
Now, it is assumed that the user voice-inputs "Tell me how to make beef bowl" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Tell me how to make beef bowl" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user's utterance information "Tell me how to make beef bowl". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "1: Tell me how to make beef bowl" from the user's utterance information.

Next, the utterance information output unit 241 acquires three communication destination identifiers. Then, the utterance information output unit 241 transmits the utterance information to the three dialogue engines 3 identified by the three communication destination identifiers. It is assumed that the three communication destination identifiers are stored in the storage unit 21.

Next, the utterance receiving unit 32 of each of the three dialogue engines 3 receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 of each of the three dialogue engines 3 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, using the learning information of the learning information storage unit 31.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from each of the three dialogue engines 3. It is assumed that the reception unit 25 first receives the automatic response information "1: I could not understand ...." from the dialogue engine "e2". Next, it is assumed that the reception unit 25 receives the automatic response information "I will introduce the material for 1: 4 servings ..." from the dialogue engine "e1". Next, the reception unit 25 received the automatic response information "1: First, water, powdered udon soup, ..." from the dialogue engine "e3". Then ... " To do.

Next, the selection unit 231 selects one automatic response information from the three automatic response information. That is, the selection unit 231 automatically responds to the dialogue engine 3 of "e1" that is first received in the automatic response information that does not include a predetermined term ("unintelligible" or "incomprehensible"). Select the information "1: 4 servings of ingredients ...".

Next, the processing unit 23 acquires the automatic information "Introducing the material for 4 servings ..." from the selected automatic response information.

Next, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information. Then, the user output unit 15 outputs the automatic information.

In addition, the automatic response information output unit 242 displays the selected automatic response information "1: 4 servings of material ...." other than the dialogue engine 3 that outputs the one automatic response information. Is transmitted to the dialogue engine 3 (dialogue engine 3 of "e2" and "e3").

Then, the learning reception unit 35 of the dialogue engine 3 of "e2" and "e3" receives the automatic response information.

Next, the learning unit 36 of the dialogue engine 3 of "e2" and "e3" performs the above-mentioned learning process using the automatic response information.

By repeating the above processing, an appropriate dialogue can be continued. In addition, the user can obtain an appropriate response. Further, the dialogue engine 3 can acquire the ability (which may be called knowledge) of another dialogue engine 3.

(Specific example 3)
Now, it is assumed that the user voice-inputs "Tell me how to make beef bowl" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Tell me how to make beef bowl" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user's utterance information "Tell me how to make beef bowl". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "1: Tell me how to make beef bowl" from the user's utterance information.

Next, the utterance information output unit 241 acquires three communication destination identifiers. Then, the utterance information output unit 241 transmits the utterance information to the three dialogue engines 3 identified by the three communication destination identifiers.

Next, the utterance receiving unit 32 of each of the three dialogue engines 3 receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 of each of the three dialogue engines 3 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, using the learning information of the learning information storage unit 31.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from each of the three dialogue engines 3.

Next, the selection unit 231 selects one automatic response information from the three automatic response information. That is, the selection unit 231 of the dialogue engine 3 of "e1" having the longest sentence length among the automatic response information that does not include a predetermined term ("unintelligible" or "incomprehensible"). Select the automatic response information "1: Introducing materials for 4 servings ....".

Next, the processing unit 23 acquires the automatic information "Introducing the material for 4 servings ..." from the selected automatic response information.

Next, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information. Then, the user output unit 15 outputs the automatic information.

In addition, the automatic response information output unit 242 displays the selected automatic response information "1: 4 servings of material ...." other than the dialogue engine 3 that outputs the one automatic response information. Is transmitted to the dialogue engine 3 (dialogue engine 3 of "e2" and "e3").

Then, the learning reception unit 35 of the dialogue engine 3 of "e2" and "e3" receives the automatic response information.

Next, the learning unit 36 of the dialogue engine 3 of "e2" and "e3" performs the above-mentioned learning process using the automatic response information.

By repeating the above processing, an appropriate dialogue can be continued. In addition, the user can obtain an appropriate response. Further, the dialogue engine 3 can acquire the ability (which may be called knowledge) of another dialogue engine 3.

In the above specific example, it is preferable to output the automatic information by voice on the user terminal 1 as the character speaks as shown in FIG.

As described above, according to the present embodiment, it is possible to effectively use a plurality of dialogue engines to continue an appropriate dialogue.

In particular, according to the present embodiment, it is possible to select appropriate automatic response information from a plurality of automatic response information transmitted from a plurality of dialogue engines and respond to the user for each dialogue. As a result, a very appropriate response can be made to the user.

The processing in this embodiment may be realized by software. Then, this software may be distributed by software download or the like. Further, this software may be recorded on a recording medium such as a CD-ROM and disseminated. This also applies to other embodiments herein. The software that realizes the dialogue control device 2 in the present embodiment is the following program. That is, this program is a program in a dialogue control device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines, and uses a computer to display user utterance information corresponding to the user's utterance. The receiving unit received from the user terminal, the utterance information output unit capable of outputting the utterance information which is the information corresponding to the user utterance information to one or more dialogue engines, and the utterance information output unit according to the output of the utterance information. The user receives an automatic response reception unit that receives automatic response information from one or more dialogue engines, and one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit. It is a program for functioning as a response transmitter to be transmitted to a terminal.

Further, in the above program, the utterance information output unit outputs the utterance information to two or more dialogue engines, and the automatic response reception unit outputs the utterance information to each of the two or more dialogue engines according to the output of the utterance information. The computer further functions as a selection unit for selecting one automatic response information from the two or more automatic response information, and the response transmission unit corresponds to the automatic response information selected by the selection unit. It is preferable that the program functions the computer so as to transmit one automatic information to the user terminal.

Further, in the above program, it is preferable that the selection unit is a program that causes the computer to function by selecting one automatic response information that matches a predetermined condition from the two or more automatic response information. is there.

Further, in the above program, the selection unit outputs from the dialogue engine having the highest priority based on the priority information among the automatic response information satisfying the predetermined conditions among the two or more automatic response information. It is preferable that the program functions the computer as the one that selects the automatic response information.

Further, the software that realizes the dialogue engine 3 in the present embodiment is the following program. That is, in this program, the recording medium accessible to the computer is provided with a learning information storage unit for storing learning information which is information for acquiring automatic response information corresponding to the spoken information, and the computer is interactively controlled. The utterance reception unit that receives utterance information from the device 2, the response acquisition unit that acquires automatic response information that is a response corresponding to the utterance information received by the utterance reception unit, and the response acquisition unit that use the learning information. A response output unit that outputs the acquired automatic response information to the dialogue control device, a learning reception unit that receives other automatic response information that is a response from another dialogue engine, and other automatic response information received by the learning reception unit. It is a program that is used to function as a learning unit that adds information to the learning information.

(Embodiment 2)
In the present embodiment, a dialogue system capable of selecting a part of dialogue engines from a plurality of dialogue engines and having a dialogue with the dialogue engines will be described. It is preferable that some dialogue engines are one, but two or more may be used.

The conceptual diagram of the dialogue system B in the present embodiment is the same as that of FIG. 1 except for the reference numerals. The dialogue system B includes one or more user terminals 1, a dialogue control device 4, and two or more dialogue engines 3. The dialogue control device 4 is a device that controls dialogue. The dialogue control device 4 is, for example, an ASP server, a cloud server, or the like. However, the type of the dialogue control device 4 does not matter.

Further, usually, one or more user terminals 1 and the dialogue control device 4 can communicate with each other via a network such as the Internet or LAN. Further, usually, the dialogue control device 4 and each of the dialogue engines 3 of 1 or more can communicate with each other via a network such as the Internet or LAN. However, the dialogue control device 4 may include one or more of one or more user terminals 1 and one or more dialogue engines 3 inside. That is, two or more of the user terminal 1, the dialogue control device 4, and the dialogue engine 3 may be physically realized by one device.

The block diagram of the dialogue system B in the present embodiment is the same as that of FIG. 2 except for the reference numerals of the dialogue control device. FIG. 10 is a block diagram of the dialogue control device 4.

The dialogue control device 4 includes a storage unit 41, a reception unit 22, a processing unit 43, an output unit 24, a reception unit 25, and a transmission unit 26. The processing unit 43 includes a selection unit 431. The output unit 24 includes an utterance information output unit 241 and an automatic information output unit 242. The reception unit 25 includes an automatic response reception unit 251.

Various types of information are stored in the storage unit 41 that constitutes the dialogue control device 4. The various information includes, for example, selection conditions of the dialogue engine 3, priority information, dialogue history, and information for communicating with each of two or more dialogue engines 3 (for example, the IP address and URL of the dialogue engine 3). , MAC address, etc.).

The processing unit 43 performs various processes. The various processes are processes performed by the selection unit 431 and the like.

The selection unit 431 selects a part of the dialogue engines 3 from the two or more dialogue engines 3. The selection unit 431 selects the dialogue engine 3 of N (N is a natural number of 2 or more) from the dialogue engine 3 of M (N is a natural number of 1 or more, M <N). The part dialogue engine 3 is a part of two or more dialogue engines 3. It is preferable that the part of the dialogue engine 3 is one dialogue engine 3.

It is preferable that the selection unit 431 selects a part of the dialogue engine 3 by using the user utterance information or the utterance information.

The selection unit 431 selects a part of the dialogue engine 3 based on the selection condition, for example. The selection condition is, for example, that the subject of dialogue (utterance) matches. The selection condition that the subject matches is, for example, to select the dialogue engine 3 in which the genre acquired from the utterance and the genre stored in association with the dialogue engine 3 match. The selection condition is, for example, that the load is small enough to satisfy a predetermined condition (for example, the load is the smallest, the number of logged-in users is equal to or less than the threshold value). The selection condition is, for example, that the response is faster (for example, the fastest response) as the predetermined condition is satisfied.

The timing at which the selection unit 431 selects the dialogue engine 3 does not matter. The selection unit 431 may select the dialogue engine 3 each time the user utterance information is received (for each dialogue). Further, the selection unit 431 may select the dialogue engine 3 at the beginning of a series of dialogues. For example, at the beginning of a series of dialogues, the selection unit 431 selects a dialogue engine 3 that matches the subject, selects a dialogue engine 3 that has the least load, or selects a dialogue engine 3 that has a fast response. You may. In order to detect the response, a command may be transmitted to the dialogue engine 3 in advance, the response may be received, and the speed of the response may be detected.

When the automatic response information is received from the plurality of dialogue engines 3, the selection unit 431 selects one automatic response information from the automatic response information. The selection is the same as the processing of the selection unit 231.

The automatic response reception unit 251 here normally receives automatic response information from one dialogue engine 3. However, the automatic response reception unit 251 may receive automatic response information from each of a plurality of dialogue engines.

The storage unit 41 is preferably a non-volatile recording medium, but can also be realized by a volatile recording medium.

The process in which the information is stored in the storage unit 41 does not matter. For example, the information may be stored in the storage unit 41 via the recording medium, the information transmitted via the communication line or the like may be stored in the storage unit 41, or The information input via the input device may be stored in the storage unit 41.

The processing unit 43 and the selection unit 431 can usually be realized from an MPU, a memory, or the like. The processing procedure of the processing unit 43 and the like is usually realized by software, and the software is recorded on a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

Next, the operation of the dialogue control device 4 will be described with reference to the flowchart of FIG. In the flowchart of FIG. 11, the same steps as the flowchart of FIG. 4 will be omitted.

(Step S1101) The selection unit 431 selects a part of the dialogue engines 3 (output destinations) from two or more dialogue engines 3.

(Step S1102) The selection unit 431 determines whether or not to select one automatic response information from the plurality of automatic response information. If one automatic response information is selected, the process goes to step S406. If one automatic response information is not selected, the process goes to step S407. Here, the selection unit 431 may determine, for example, whether or not two or more dialogue engines 3 have been selected in step S1101. The selection unit 431 may determine, for example, whether or not two or more automatic response information has been accepted.

In the flowchart of FIG. 11, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Hereinafter, the specific operation of the dialogue system B in the present embodiment will be described. In this specific example, it is assumed that the dialogue system B includes three dialogue engines. The conceptual diagram of the dialogue system A is FIG.

It is assumed that the learning information storage unit 31 constituting one of the three dialogue engines 3 (for example, the dialogue engine "e1") stores the learning information management table shown in FIG. 7. .. Further, it is assumed that the learning information storage unit 31 of the dialogue engine "e1" stores the Foodtable shown in FIG. Further, it is assumed that the learning information storage unit 31 of the dialogue engine "e1" has a food dictionary for storing the dish name and a crops dictionary for storing the crop name. The food dictionary stores a large number of dish names such as "Katsudon, Gyudon, Oyakodon, ...". In addition, a large number of crop names such as "rice, eggplant, carrot, ..." are stored in the crops dictionary.

In such a situation, the following two specific examples will be described. Specific example 1 is a case where one dialogue engine is selected. Specific example 2 is a case where two or more dialogue engines are selected and two or more automatic response information is selected. As a method for selecting two or more automatic response information, various algorithms may be used as described in the first embodiment.

(Specific example 1)
Now, it is assumed that the storage unit 21 of the dialogue control device 2 stores the selection condition management table shown in FIG. The selection condition management table is a table that manages the conditions for selecting the dialogue engine. The selection condition management table has three records having "interaction engine identifier", "genre", and "notification destination identifier". A "genre" is a genre that each dialogue engine can respond to.

In such a situation, it is assumed that the user voice-inputs "Tell me how to make beef bowl" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Tell me how to make beef bowl" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user's utterance information "Tell me how to make beef bowl". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information "Tell me how to make beef bowl" from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "1: Tell me how to make beef bowl" from the user's utterance information.

Next, the selection unit 431 selects one dialogue engine 3 from two or more dialogue engines 3. Here, the selection unit 431 performs natural language processing on "Tell me how to make beef bowl" and acquires the genre "cooking". Then, the selection unit 431 selects the dialogue engine "e2" that is paired with the genre "cooking". In addition, the selection unit 431 performs morphological analysis of, for example, "teach me how to make beef bowl", acquires the independent words "gyudon", "how to make", and "teach", and machine learning (for example, deep layer) from the independent words. Acquire the genre "cooking" by learning, decision tree, SVM). Since the technique of classifying sentences by machine learning and determining the genre (which may be called a class) of the sentence is a known technique, detailed description thereof will be omitted. In addition, the selection unit 431 performs morphological analysis of, for example, "tell me how to make beef bowl", acquires the independent words "gyudon", "how to make", and "teach", and performs a DB search using the independent word as a key. , Get the genre "cooking" from the DB. In such a case, it is assumed that the record corresponding to the independent word and the genre is stored in the DB. The genre input by the user may be received from the user terminal 1, the receiving unit 22 may receive the genre, and the selection unit 431 may acquire the genre "cooking" using the genre.

Then, the utterance information output unit 241 acquires the communication destination identifier of the dialogue engine "e2" from the table of FIG. 12, and informs the dialogue engine 3 identified by the communication destination identifier of the utterance information "1: How to make beef bowl". Tell me "is sent.

Next, the utterance receiving unit 32 of the dialogue engine "e2" receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 provides the automatic response information "1: 4 servings of material ...", which is a response corresponding to the utterance information received by the utterance reception unit 32, to the learning information storage unit 31. Acquired using the learning information of. Then, the response output unit 34 transmits the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from the dialogue engine "e2".

Next, the processing unit 23 acquires the automatic information "Introducing the material for 4 servings ..." from the selected automatic response information.

Next, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information. Then, the user output unit 15 outputs the automatic information.

In addition, the automatic response information output unit 242 provides the selected automatic response information "1: Introduce the ingredients for 4 servings ..." and the utterance information "1: Tell me how to make beef bowl". , The automatic response information of the one is transmitted to the dialogue engine 3 (the dialogue engine 3 of "e1" and "e3") other than the dialogue engine 3 that outputs the information.

Then, the learning reception unit 35 of the dialogue engine 3 of "e1" and "e3" receives the automatic response information and the like.

Next, the learning unit 36 of the dialogue engine 3 of "e1" and "e3" configures information to be added to the learning information by using the automatic response information received by the learning reception unit 35. In other words, the learning department 36 learns the combination of the content of the utterance information "Tell me how to make beef bowl" and the content of the self-response information "Introducing the ingredients for 4 servings ...." Add to the information management table.

The learning unit 36 replaces the nouns or noun phrases "beef bowl" and "how to make" with variables in the content of the utterance information "tell me how to make beef bowl", and replaces "{variable 2} of {variable 1}" with variables. Change to the pattern "Tell me" and replace the noun or noun phrase "4 servings" and "material" with variables in "Introducing materials for 4 servings ...." You may change the pattern to "Introducing variable 4} ...." and add a set of two patterns to the learning information management table in Fig. 8.

That is, in this specific example, it is sufficient that the learning information is enriched by using the automatic response information input from the other dialogue engine 3, and the structure of the learning information does not matter.

By repeating the above processing, an appropriate dialogue can be continued. In addition, the user can obtain an appropriate response. Further, the dialogue engine 3 can acquire the ability (which may be called knowledge) of another dialogue engine 3.

(Specific example 2)
Now, it is assumed that the storage unit 21 of the dialogue control device 2 stores the selection condition management table shown in FIG. Further, the storage unit 21 stores the priority information management table shown in FIG.

In such a situation, it is assumed that the user voice-inputs "Tell me a hotel where futsal can be played in Kyoto" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the story. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user utterance information "Tell me a hotel where futsal can be played in Kyoto". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 2 receives the user utterance information from the user terminal 1.

Then, the processing unit 23 acquires the utterance information "1: Tell me the hotel where futsal can be played in Kyoto" from the user's utterance information.

Next, the selection unit 431 selects one or two or more dialogue engines 3 from two or more dialogue engines 3. Here, the selection unit 431 performs natural language processing on "Tell me a hotel where you can play futsal in Kyoto" and acquires the genre "Travel, sports". Then, the selection unit 431 selects the dialogue engines "e1" and "e3" that are paired with the genre "travel" or "sports".

Then, the utterance information output unit 241 acquires the communication destination identifiers of the dialogue engines "e1" and "e3" from the table of FIG. 12, and supplies the utterance information "1:" to the two dialogue engines 3 identified by the communication destination identifiers. Tell me which hotels you can play futsal in Kyoto. "

Next, the utterance receiving unit 32 of each of the dialogue engines of "e1" and "e3" receives the utterance information from the dialogue control device 2. Then, the response acquisition unit 33 of each of the two dialogue engines 3 acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit 32, using the learning information of the learning information storage unit 31. Next, the response output unit 34 of each of the two dialogue engines 3 transmits the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2.

Next, the reception unit 25 of the dialogue control device 2 receives the automatic response information from each of the two dialogue engines 3.

Next, the selection unit 431 selects one automatic response information from the two automatic response information. Such a selection method has been described in the first embodiment. Further, it is assumed that the selection unit 431 has selected the automatic response information from the dialogue engine "e1".

Next, the processing unit 23 acquires automatic information from the selected automatic response information. Then, the response transmission unit 261 transmits the automatic information to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the dynamic information. Then, the user output unit 15 outputs the automatic information.

Further, the automatic response information output unit 242 transfers the selected automatic response information to the dialogue engine 3 (dialogue engine 3 of "e2" and "e3") other than the dialogue engine 3 that outputs the one automatic response information. Send.

Then, the learning reception unit 35 of the dialogue engine 3 of "e2" and "e3" receives the automatic response information.

Next, the learning unit 36 of the dialogue engine 3 of the “e2” and “e3” performs the above-mentioned learning process using the automatic response information received by the learning reception unit 35.

By repeating the above processing, an appropriate dialogue can be continued. In addition, the user can obtain an appropriate response. Further, the dialogue engine 3 can acquire the ability (which may be called knowledge) of another dialogue engine 3.

As described above, according to the present embodiment, it is possible to effectively use a plurality of dialogue engines to continue an appropriate dialogue. In particular, according to the present embodiment, by selecting a dialogue engine from a plurality of dialogue engines, it is possible to provide an appropriate response to the user with a small load.

The software that realizes the dialogue control device 4 in this embodiment is the following program. That is, this program is a program in a dialogue control device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines, and uses a computer to display user utterance information corresponding to the user's utterance. The receiving unit received from the user terminal, the utterance information output unit capable of outputting the utterance information which is the information corresponding to the user utterance information to one or more dialogue engines, and the utterance information output unit according to the output of the utterance information. The user receives an automatic response reception unit that receives automatic response information from one or more dialogue engines, and one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit. It is a program for functioning as a response transmitter to be transmitted to a terminal.

Further, in the above program, the computer is further functioned as a selection unit for selecting a part of the dialogue engines from the two or more dialogue engines, and the speech information output unit selects the speech information. Output to the dialogue engine of the unit, the automatic response receiving unit receives automatic response information from some of the dialogue engines in response to the output of the spoken information, and the response transmitting unit receives the automatic response receiving unit. It is preferable that the program functions the computer so as to transmit one automatic information corresponding to the automatic response information to the user terminal.

Further, in the above program, it is preferable that the selection unit is a program for operating a computer by using the user utterance information or the utterance information to select one dialogue engine.

(Embodiment 3)
In the present embodiment, a dialogue system including a dialogue control device having an automatic mode that adopts a response from a plurality of dialogue engines and a manual mode that adopts a response of an administrator will be described.

Further, in the present embodiment, a dialogue system that automatically determines or automatically changes the mode will be described. In the present embodiment, the failure of the dialogue or the like is automatically detected, and the mode is automatically changed.

Further, in the present embodiment, a dialogue system that changes the mode according to an instruction from the administrator will be described.

Further, in the present embodiment, a dialogue system that outputs a response from the dialogue engine to the management terminal in the case of the automatic mode will be described.

Further, in the present embodiment, a dialogue system that provides a response from the administrator to the dialogue engine in the case of the manual mode will be described.

FIG. 13 is a conceptual diagram of the dialogue system C in the present embodiment. The dialogue system A includes one or more user terminals 1, a dialogue control device 5, two or more dialogue engines 6, and one or more management terminals 7. The dialogue control device 5 is a device that controls dialogue. The dialogue control device 5 is, for example, an ASP server, a cloud server, or the like. However, the type of the dialogue control device 5 does not matter. The dialogue engine 6 is a device that receives information on utterances from the user and returns a response. The dialogue engine 6 is, for example, an ASP server, a cloud server, or the like. However, the type of the dialogue engine 6 does not matter. Further, the dialogue engine 6 receives the utterance information and outputs the automatic response information. The dialogue engine 6 may be a known dialogue engine. The management terminal 7 is, for example, a so-called personal computer, a tablet terminal, a smartphone, or the like, and the type thereof does not matter. The management terminal 7 is a terminal used by the administrator. An administrator is the administrator of a conversation, who interacts with the user in manual mode. It can be said that the manager is a person who supports the dialogue.

Further, usually, one or more user terminals 1 and the dialogue control device 5 can communicate with each other via a network such as the Internet or a LAN. Further, normally, the dialogue control device 5 and the dialogue engine 6 can communicate with each other via a network such as the Internet or a LAN. Further, normally, the dialogue control device 5 and one or more management terminals 7 can communicate with each other via a network such as the Internet or a LAN. However, the dialogue control device 5 may include one or more of the dialogue engine 6 and the management terminal 7 inside. That is, two or more of the dialogue control device 5, the dialogue engine 6, and the management terminal 7 may be physically realized by one device. Further, the dialogue control device 5 may include one or more of the user terminal 1 and the dialogue engine 6 inside. That is, two or more of the dialogue control device 5, the user terminal 1, and the dialogue engine 6 may be physically realized by one device.

FIG. 14 is a block diagram of the dialogue system C according to the present embodiment. Further, FIG. 15 is a block diagram of the dialogue control device 5.

The dialogue control device 5 includes a storage unit 51, a reception unit 22, a processing unit 53, an output unit 54, a reception unit 55, and a transmission unit 56. The storage unit 51 includes a mode storage unit 511. The processing unit 53 includes a determination unit 531, a determination unit 532, a mode change unit 533, and a selection unit 534. The output unit 54 includes an utterance information output unit 541, an automatic information output unit 542, and a manual information output unit 543. The reception unit 55 includes an automatic response reception unit 551, a manual response reception unit 552, and a change instruction reception unit 553. The transmission unit 56 includes a response transmission unit 561.

The dialogue engine 6 includes a learning information storage unit 31, an utterance reception unit 32, a response acquisition unit 33, a response output unit 34, a learning reception unit 65, and a learning unit 66.

The management terminal 7 includes a management information reception unit 71, a management processing unit 72, an administrator output unit 73, a management instruction reception unit 74, and a management output unit 75.

Various types of information are stored in the storage unit 51 that constitutes the dialogue control device 5. The various types of information are, for example, mode information described later, selection conditions for automatic response information, selection conditions for the dialogue engine 3, priority information, and dialogue history.

Mode information is stored in the mode storage unit 511. The mode information is information indicating whether the interaction with the user is automatic or manual. The mode information usually includes an automatic mode in which any of the dialogue engines 6 is adopted, or a manual mode in which the management terminal 7 is adopted. That is, the mode information can be at least either an automatic mode or a manual mode. However, the mode information may be a mode other than the automatic mode and the manual mode. The modes other than the automatic mode and the manual mode are, for example, a stop mode that does not operate. Further, adopting the dialogue engine 6 means that the response from the dialogue engine 6 is transmitted to the user terminal 1. Further, adopting the management terminal 7 means that the response from the management terminal 7 is transmitted to the user terminal 1.

The processing unit 53 performs various processes. The various processes are processes performed by the determination unit 531, the determination unit 532, the mode change unit 533, the selection unit 534, and the like.

The various processes are, for example, processes for acquiring utterance information from user utterance information. The processing unit 53 generates an utterance identifier in response to receiving user utterance information, for example. Then, the processing unit 53 adds a utterance identifier to the user utterance information, for example, to form the utterance information. The utterance identifier is an identifier of user utterance information and is unique information. Further, the processing unit 53 increments, for example, the recently generated utterance identifier by 1, and generates a unique utterance identifier.

Further, the processing unit 53 may acquire the automatic information or the manual information from the automatic response information or the manual response information, for example. For example, the processing unit 53 may generate automatic information which is the voice data from the automatic response information of the text. The processing unit 53 may generate the manual information which is the voice data from the manual response information of the text, for example. Further, the processing unit 53 may acquire the automatic information of the data structure to be transmitted or the manual information of the data structure to be transmitted from, for example, the automatic response information or the manual response information. In such a case, the processing unit 53 changes the data structure of the automatic response information or the manual response information, for example, and acquires the automatic information or the manual information. In addition, for example, when the automatic response information and the automatic information are the same information, the processing unit 53 simply acquires the automatic response information received by the reception unit 55. Further, for example, when the manual response information and the manual information are the same information, the processing unit 53 simply acquires the manual response information received by the reception unit 55.

The determination unit 531 determines whether to adopt the dialogue engine 6 or the management terminal 7. The adoption of the dialogue engine 6 is the adoption of the response of the dialogue engine 6 and is to transmit the automatic information which is the information for the automatic response information which is the response of the dialogue engine 6 to the user terminal 1. Further, the adoption of the management terminal 7 is the adoption of the response from the management terminal 7, and the manual information which is the information for the manual response information which is the response of the management terminal 7 is transmitted to the user terminal 1.

The determination unit 531 acquires the mode information of the mode storage unit 511. Then, the determination unit 531 determines whether to adopt the dialogue engine 6 or the management terminal 7 according to the mode information of the mode storage unit 511. This mode information is, for example, an automatic mode indicating that the dialogue engine 6 is adopted, or a manual mode indicating that the management terminal 7 is adopted.

The determination process of the determination unit 531 may be performed by reading the mode information of the mode storage unit 511.

Further, the determination unit 531 may determine the output destination of the utterance information. The determination unit 531 determines, for example, whether the output destination of the utterance information is the dialogue engine 6 or the management terminal 7 according to the mode information of the mode storage unit 511. Further, the determination unit 531 may always set the output destination of the utterance information to both the dialogue engine 6 and the management terminal 7.

The determination unit 532 determines whether or not a predetermined condition is satisfied. Here, the predetermined condition is a condition for changing the mode information. The conditions for changing the mode information include the first condition for changing from the automatic mode to the manual mode and the second condition for changing from the manual mode to the automatic mode.

The first condition is, for example, a condition for identifying that the dialogue is broken, an instruction from the management terminal 7 is accepted, and the like. The condition for identifying that the dialogue is broken is, for example, the condition for detecting the breakdown of the dialogue by using one or more automatic response information which is the response from the dialogue engine 6, from the user terminal 1. It is a condition for detecting the failure of the dialogue using one or more user speech information of the above, and a condition for detecting the failure of the dialogue based on the input from the management terminal 7. The conditions for identifying that the dialogue is broken include, for example, predetermined terms (for example, "incomprehensible" and "incomprehensible") in the automatic response information which is the response from the dialogue engine 6. That, the automatic response information including the predetermined term is received more than the threshold or more times than the threshold within the predetermined period, and the received two or more consecutive user utterance information are the same or similar to the threshold or more. It means that the user's utterance information includes predetermined terms (for example, "what are you saying" and "no good"). Further, the instructions from the management terminal 7 are, for example, an instruction to change the mode information and an instruction to specify the mode information. The failure of the dialogue is the failure of the dialogue between the user and the dialogue engine 6. Further, the predetermined conditions are stored in, for example, the storage unit 51.

The second condition is, for example, that the response from the management terminal 7 has been received a number of times equal to or greater than the threshold value, that the instruction from the management terminal 7 has been received, and the like.

The mode change unit 533 changes the mode information when the determination unit 532 determines that the predetermined condition is satisfied. Changing the mode information is usually writing the mode information to the mode storage unit 211. The mode change unit 533 changes the mode information of the mode storage unit 511 in response to the change instruction.

The selection unit 534 selects one automatic response information from, for example, two or more automatic response information. The selection unit 534 selects one automatic response information that matches a predetermined condition from two or more automatic response information. The selection unit 534 selects the automatic response information output from the dialogue engine 6 having the highest priority based on the priority information among the automatic response information satisfying the predetermined conditions among the two or more automatic response information. To do. The selection unit 534 selects, for example, one dialogue engine 6 from two or more dialogue engines 6. The selection unit 534 selects one dialogue engine 6 by using the user utterance information or the utterance information. The selection unit 534 may perform the same processing as the selection unit 231 or the selection unit 431.

The output unit 54 outputs various information. The output here is usually a transmission. However, the output means that the information should be passed to the output destination.

The utterance information output unit 541 outputs utterance information, which is information corresponding to the user utterance information, to at least one of the dialogue engine 6 and the management terminal 7. The utterance information output unit 541 may output the utterance information to the dialogue engine 6 and the management terminal 7 regardless of the mode information. Here, the utterance information may be the same information as the user utterance information. Further, the utterance information may be a text obtained by voice-recognizing the user utterance information. It is preferable that the utterance information is information to which the utterance identifier is added.

The utterance information output unit 541 can output the utterance information to the management terminal 7 as described above. When the determination unit 531 decides to adopt the management terminal 7, the utterance information output unit 541 outputs the utterance information to the management terminal 7. Even when the determination unit 531 decides to adopt the management terminal 7, the utterance information may be output to one or more dialogue engines 6.

The utterance information output unit 541 can output utterance information, which is information corresponding to the user utterance information, to one or more dialogue engines 6. The utterance information output unit 541 may output the utterance information to two or more dialogue engines 6. The utterance information output unit 541 may output the utterance information to one dialogue engine 6 selected by the selection unit 534.

The automatic information output unit 542 outputs one automatic information to the management terminal 7. The automatic information output unit 542 outputs the automatic information corresponding to the automatic response information received by the automatic response reception unit 551 to the management terminal 7.

The manual information output unit 543 outputs the manual information to one or more dialogue engines 6. The manual information output unit 543 outputs the manual information corresponding to the manual response information received by the manual response reception unit 552 to one or more dialogue engines 6.

The reception unit 55 receives various information, instructions, and the like. Acceptance here is usually reception. However, the reception may be as long as the information etc. is received.

The automatic response reception unit 551 receives automatic response information which is a response from each of the one or more dialogue engines 6 in response to the output of the utterance information to one or more dialogue engines 6. Such reception is, for example, reception.

The manual response reception unit 552 receives the manual response information, which is a response from the management terminal 7, in response to the output of the utterance information to the management terminal 7. Such reception is, for example, reception.

The change instruction receiving unit 553 receives a change instruction, which is an instruction for changing the mode, from the management terminal 7. Such reception is, for example, reception. Further, this change instruction may be a change from the automatic mode to the manual mode, or may be a change from the manual mode to the automatic mode.

The transmission unit 56 transmits various types of information. The various types of information include, for example, manual information, automatic information, and the like.

The response transmission unit 561 transmits the automatic information which is the information corresponding to the automatic response information or the manual information which is the information corresponding to the manual response information to the user terminal 1.

The response transmission unit 561 may transmit automatic information to the user terminal 1 when the decision of the determination unit 531 determines that the dialogue engine 6 is adopted. Further, the response transmission unit 561 may transmit manual information to the user terminal 1 when the decision of the determination unit 531 determines that the management terminal 7 is adopted. Here, the automatic information may be the same information as the automatic response information. The contents of the automatic information and the automatic response information are usually the same. The automatic information is, for example, voice data obtained by converting automatic response information into voice. The manual information may be the same information as the manual response information. The contents of the manual information and the manual response information are usually the same. The manual information is, for example, voice data obtained by converting manual response information into voice.

The response transmission unit 561 may transmit one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit 551 to the user terminal 1. The response transmission unit 561 may transmit one automatic information corresponding to the automatic response information selected by the selection unit 534 to the user terminal 1. The response transmission unit 561 may transmit one automatic information corresponding to the automatic response information received by the automatic response reception unit 551 to the user terminal 1.

The learning reception unit 65 constituting the dialogue engine 6 receives the manual response information which is the response from the management terminal 7. Further, the learning reception unit 65 receives automatic response information which is a response from another dialogue engine 6. The acceptance of the manual response information or the automatic response information here is intended to accept the manual information or the automatic information.

The learning unit 66 performs learning processing using the manual response information received by the learning reception unit 35. In addition, the learning unit 66 performs learning processing using the automatic response information received by the learning reception unit 35. The learning process using the automatic response information has been described above.

The learning process is, for example, adding information to the learning information of the learning information storage unit 31 by using the manual response information. The learning process is, for example, changing the learning information of the learning information storage unit 31 by using the manual response information. The process of adding information to the learning information includes a process of adding new learning information to the learning information storage unit 31.

The learning unit 66 adds, for example, learning information, which is a set of manual response information and utterance information corresponding to the manual response information, to the learning information storage unit 31.

For example, the learning unit 66 converts the information of the set of the manual response information and the utterance information corresponding to the manual response information into the learning information which is the state transition information, and the node (state) of the automatic response information before the utterance information. As the previous state, the utterance information node is added, and the manual response information node is added to the utterance information node.

For example, the learning unit 66 constitutes a pattern of manual response information from manual response information, and constitutes a pattern of utterance information from utterance information corresponding to the manual response information, and sets a pattern of utterance information and a pattern of response. The learning information is added to the learning information storage unit 31. The response pattern is the pattern of automatic response information.

The learning unit 66, for example, constructs a pattern of manual response information from manual response information, and configures a pattern of utterance information from utterance information corresponding to the manual response information, and constitutes a pattern of utterance information that matches the set of the patterns. The learning information of the learning information storage unit 31 is updated so as to increase the appearance frequency or appearance probability of the pattern and the pattern set of the automatic response information. In such a case, the learning information has a set of a pattern of utterance information and a pattern of automatic response information, and an appearance frequency or appearance probability.

The learning unit 66 performs the above learning process using, for example, the utterance information and the manual response information in which the utterance identifier corresponding to the utterance information and the utterance identifier corresponding to the manual response information are the same.

In addition, there may be various learning processes using the manual response information performed by the learning unit 66. The learning process performed by the learning unit 66 may use another known learning processing algorithm, except that the manual response information or the automatic response information from another dialogue engine 6 is used.

Various information is stored in the management storage unit 70 that constitutes the management terminal 7. Various types of information include, for example, mode information, utterance information, automatic information, manual response information, and the like.

The management information reception unit 71 receives manual response information from the administrator. The reception here may be reception by voice or reception from an input means such as a keyboard or a mouse.

It is preferable that the management information receiving unit 71 receives the manual response information from the administrator by voice. Further, the management information reception unit 71 may receive manual response information from the administrator from a keyboard, a mouse, a touch panel, a menu screen, or the like. It is preferable that the management information reception unit 71 receives one manual response information by voice or by an input means such as a keyboard. That is, it is preferable that the management information receiving unit 71 includes both a voice receiving means and an input means such as a keyboard, a mouse, a touch panel, or a menu screen, and receives one manual response information by any means. Is.

The management information reception unit 71 receives automatic information from the dialogue control device 5. Further, the management information reception unit 71 receives the utterance information from the dialogue control device 5. The reception here is usually reception, but any process may be used as long as the information is passed to the management terminal 7.

The management processing unit 72 performs various processes. The various processes are, for example, a process of forming a data structure for outputting automatic information or utterance information received by the management information receiving unit 71, or a process of configuring an output screen of such information.

The administrator output unit 73 performs various processes. The various processes are, for example, a process of forming a data structure for outputting automatic information or utterance information received by the management information receiving unit 71, or a process of configuring an output screen of such information.

The management instruction receiving unit 74 receives various instructions from the administrator. The various instructions are, for example, change instructions that are instructions for changing the mode. The input means for various instructions may be any means such as a keyboard, a touch panel, a mouse, and a menu screen. The management instruction receiving unit 74 can be realized by a device driver of an input means such as a keyboard, control software of a menu screen, or the like.

The management output unit 75 outputs various information and instructions to the dialogue control device 5. Various types of information and instructions are, for example, change instructions, mode information, and the like. The output here is usually transmission, but information or the like may be passed to the dialogue control device 2.

The storage unit 51, the mode storage unit 511, and the management storage unit 70 are preferably non-volatile recording media, but can also be realized by volatile recording media.

The process of storing information in the storage unit 51 or the like does not matter. For example, the information may be stored in the storage unit 51 or the like via the recording medium, or the information transmitted via the communication line or the like may be stored in the storage unit 51 or the like. Alternatively, the information input via the input device may be stored in the storage unit 51.

The processing unit 53, the determination unit 531, the determination unit 532, the mode change unit 533, the selection unit 534, the response acquisition unit 63, the learning unit 66, and the management processing unit 72 can usually be realized from an MPU, a memory, or the like. The processing procedure of the processing unit 53 is usually realized by software, and the software is recorded in a recording medium such as a ROM. However, it may be realized by hardware (dedicated circuit).

The output unit 54, the utterance information output unit 541, the automatic information output unit 542, the manual information output unit 543, the transmission unit 56, the response transmission unit 561, the response output unit 64, and the management output unit 75 are usually wireless or wired communication. It is realized by means, but it may be realized by broadcasting means or the like.

The output unit 54, the utterance information output unit 541, the automatic information output unit 542, the manual information output unit 543, and the response output unit 64 may be realized by output means for outputting information. The output of information here is, for example, the delivery of information by a program.

The reception unit 55, the automatic response reception unit 551, the manual response reception unit 552, the change instruction reception unit 553, the utterance reception unit 62, the learning reception unit 65, and the management information reception unit 71 are usually realized by wireless or wired communication means. However, it may be realized by a broadcast receiving means or the like.

The reception unit 55, the automatic response reception unit 551, the manual response reception unit 552, the change instruction reception unit 553, the utterance reception unit 62, the learning reception unit 65, and the management information reception unit 71 are realized by the reception means for receiving information. You may. The acceptance of information here is, for example, the delivery of information by a program.

The administrator output unit 73 may or may not include an output device such as a display or a speaker. The administrator output unit 73 can be realized by the driver software of the output device, the driver software of the output device, the output device, or the like.

Next, the operation of the dialogue system C will be described. First, the operation of the dialogue control device 5 will be described with reference to the flowchart of FIG. In the flowchart of FIG. 16, the description of the same steps as the flowchart of FIG. 4 will be omitted.

(Step S1601) The determination unit 531 and / and the selection unit 534 determine the output destination of the utterance information. The determination unit 531 determines, for example, whether the output destination is the dialogue engine 3 or the management terminal 7 according to the mode information of the mode storage unit 211. The determination unit 531 may always set the output destination of the utterance information to both one or more dialogue engines 6 and the management terminal 7. When the mode information is the automatic mode, the selection unit 534 selects one or more dialogue engines 6 out of two or more dialogue engines 6. As such a selection method, any of the methods described in the first and second embodiments can be used. The selection unit 534 may determine all two or more dialogue engines 6 as output destinations of utterance information.

(Step S1602) The utterance information output unit 241 outputs the utterance information acquired in step S402 to the output destination determined in step S1601.

(Step S1603) The reception unit 55 receives one or more automatic response information or manual response information according to the output of the utterance information. The reception unit 55 waits until it receives the automatic response information or the manual response information.

(Step S1604) The selection unit 534 determines whether or not to select one automatic response information from the plurality of automatic response information. If one automatic response information is selected, the process goes to step S1605. If one automatic response information is not selected, the process goes to step S1606. Here, the selection unit 534 may determine, for example, whether or not two or more dialogue engines 3 have been selected in step S1601. The selection unit 534 may determine, for example, whether or not two or more automatic response information has been accepted.

(Step S1605) The selection unit 534 selects one automatic response information from two or more automatic response information.

(Step S1606) The processing unit 53 acquires the automatic information or the manual information from the received automatic response information or the manual response information. When both the automatic response information and the manual response information are received, the processing unit 23 acquires the automatic information and the manual information from the automatic response information and the dynamic response information.

(Step S1607) The processing unit 53 determines whether or not to output the automatic information to the management terminal 7. If it is output to the management terminal 7, it goes to step S1608, and if it is not output, it goes to step S1609. For example, a flag indicating whether or not to output automatic information to the management terminal 7 is managed by the storage unit 51, and the processing unit 53 reads the flag from the storage unit 51 and determines. Further, the processing unit 53 may always output the automatic information to the management terminal 7, or may not always output the automatic information to the management terminal 7.

(Step S1608) The automatic information output unit 242 outputs the automatic information acquired in step S1606 to the management terminal 7.

(Step S1609) The processing unit 53 determines whether or not to output the manual information to the dialogue engine 6. If it is output to the dialogue engine 6, it goes to step S1610, and if it is not output, it goes to step S1611. For example, a flag indicating whether or not to output manual information to the dialogue engine 6 is managed by the storage unit 51, and the processing unit 53 reads such a flag from the storage unit 51 and determines. Further, the processing unit 53 may always output the manual information to the dialogue engine 6, or may not always output the manual information to the dialogue engine 6.

(Step S1610) The manual information output unit 543 outputs the manual information acquired in step S1606 to one or more dialogue engines 6. The manual information output unit 543 may output the manual information to all the plurality of dialogue engines 6 in which the dialogue control device 5 can be used, or may output the manual information to only some of the dialogue engines 6. good. It is preferable that the dialogue engine 6 to which the manual information is output is a dialogue engine having a learning function.

(Step S1611) The determination unit 531 determines whether to adopt the dialogue engine 6 or the management terminal 7. Then, the determination unit 531 acquires the automatic information when it is determined that the dialogue engine 6 is adopted, and acquires the manual information when it is determined that the management terminal 7 is adopted.

(Step S1612) The response transmission unit 561 transmits the automatic information or manual information acquired in step S1611 to the user terminal 1.

(Step S1613) The determination unit 532 determines whether or not the condition for changing the mode information is satisfied.

(Step S1614) If the determination result in step S1613 is the determination result that the condition is satisfied, the process goes to step S1615, and if the determination result is that the condition is not satisfied, the process returns to step S1601.

(Step S1615) The mode change unit 533 changes the mode information of the mode storage unit 511. Return to step S1601.

(Step S1616) The processing unit 53 determines whether or not the change in step S1615 is a change to the manual mode. If it is a change to the manual mode, it goes to step S1617, and if it is not a change to the manual mode, it goes to step S1618.

(Step S1617) The output unit 54 outputs the latest utterance information to the management terminal 7. Return to step S1605.

(Step S1618) The output unit 54 outputs the latest utterance information to the dialogue engine 6. Return to step S1605.

(Step S1619) The change instruction receiving unit 553 determines whether or not the change instruction has been received from the management terminal 7. If the change instruction is accepted, the process proceeds to step S1620, and if the change instruction is not accepted, the process returns to step S1601.

(Step S1620) The mode change unit 533 changes the mode information of the mode storage unit 511. Return to step S1601.

In the flowchart of FIG. 16, the timing of the determination process of the determination unit 532 in step S1613 does not matter.

Further, it goes without saying that in the flowchart of FIG. 16, the order of various processes such as the order of transmission of automatic information and manual information and the order of determination of whether or not to transmit information does not matter.

Further, in the flowchart of FIG. 16, if the automatic response information is not received in step S1605, the automatic information is not output. If the manual response information is not accepted in step S1605, the manual information is not output.

Further, in the flowchart of FIG. 16, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Next, the operation of the dialogue engine 6 will be described with reference to the flowchart of FIG.

(Step S1701) The utterance reception unit 32 determines whether or not the utterance information has been received from the dialogue control device 2. If the utterance information is accepted, the process goes to step S1702, and if the utterance information is not accepted, the process goes to step S1704.

(Step S1702) The response acquisition unit 33 acquires the automatic response information corresponding to the utterance information received in step S1701 by using the learning information of the learning information storage unit 31.

(Step S1703) The response output unit 34 outputs the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 2.

(Step S1704) The learning reception unit 65 determines whether or not the learning target has been accepted. If the learning target is accepted, the process proceeds to step S1705, and if the learning target is not accepted, the process returns to step S1701. The learning target is manual response information or automatic response information that is the output of another dialogue engine 6.

(Step S1705) The learning unit 66 configures information to be added to the learning information by using the learning target received by the learning reception unit 65.

(Step S1706) The learning unit 66 adds the information configured in step S1705 to the learning information storage unit 31. Return to step S1701.

In the flowchart of FIG. 5, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Next, the operation of the dialogue engine 6 will be described with reference to the flowchart of FIG.

(Step S1701) The utterance reception unit 32 determines whether or not the utterance information has been received from the dialogue control device 5. If the utterance information is accepted, the process goes to step S1702, and if the utterance information is not accepted, the process goes to step S1704.

(Step S1702) The response acquisition unit 33 acquires the automatic response information corresponding to the utterance information received in step S1701 by using the learning information of the learning information storage unit 31.

(Step S1703) The response output unit 34 outputs the automatic response information acquired by the response acquisition unit 33 to the dialogue control device 5.

(Step S1704) The learning reception unit 65 determines whether or not the automatic response information or the manual response information to be learned has been received. If the automatic response information or the manual response information is accepted, the process proceeds to step S1705, and if the automatic response information is not accepted, the process returns to step S1701.

(Step S1705) The learning unit 66 configures information to be added to the learning information by using the automatic response information or the manual response information received by the learning reception unit 65.

(Step S1706) The learning unit 66 adds the information configured in step S1705 to the learning information storage unit 31. Return to step S1701.

In the flowchart of FIG. 17, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Further, in the flowchart of FIG. 17, the above-mentioned learning algorithm and known algorithm can be adopted for the learning process itself.

Next, the operation of the management terminal 7 will be described with reference to the flowchart of FIG.

(Step S1801) The management information reception unit 71 determines whether or not the utterance information has been received from the dialogue control device 5. If the utterance information is accepted, the process goes to step S1802, and if the utterance information is not accepted, the process goes to step S1806.

(Step S1802) The administrator output unit 73 outputs the utterance information received in step S1801 or the utterance information configured for output by the management processing unit 72.

(Step S1803) The management processing unit 72 determines whether or not to output the manual response information corresponding to the output of the utterance information. If the manual response information is output, the process proceeds to step S1804, and if the manual response information is not output, the process returns to step S1801. When the manual response information is output, the mode information is the manual mode. For example, when the mode information of the management storage unit 70 is "manual mode", the management processing unit 72 determines that the manual response information is output.

(Step S1804) The management information reception unit 71 determines whether or not the manual response information has been received. If the manual response information is accepted, the process proceeds to step S1805, and if the manual response information is not accepted, the process returns to step S1804.

(Step S1805) The administrator output unit 73 outputs the manual response information received in step S1804 to the dialogue control device 5. Return to step S1801.

(Step S1806) The management information receiving unit 71 determines whether or not the automatic information has been received from the dialogue control device 6. If the automatic information is accepted, the process goes to step S1807, and if not, the process goes to step S1808.

(Step S1807) The administrator output unit 73 outputs the automatic information received in step S1806 or the automatic information configured for output by the management processing unit 72. Return to step S1801.

(Step S1808) The management instruction receiving unit 74 determines whether or not the mode change instruction has been received from the administrator. If the change instruction is accepted, the process proceeds to step S1809, and if the change instruction is not accepted, the process returns to step S1801.

(Step S1809) The administrator output unit 75 outputs the change instruction received in step S1808 to the dialogue control device 5. Return to step S1801.

In the flowchart of FIG. 6, the process ends when the power is turned off or an interrupt for the end of the process occurs.

Hereinafter, the specific operation of the dialogue system C in the present embodiment will be described. The conceptual diagram of the dialogue system C is FIG.

Now, it is assumed that the storage unit 51 of the dialogue control device 5 stores a failure determination condition management table, which is information indicating a condition for identifying that the dialogue has failed (see FIG. 19). The bankruptcy judgment condition management table manages one or more records having "ID" and "bankruptcy condition". The "bankruptcy condition" has "information" and "condition". The "ID" is information that identifies a record. "Information" is information used to determine the failure of a dialogue. "Information" is user utterance information or automatic response information. The utterance information may be used instead of the user utterance information, or the automatic information may be used instead of the automatic response information. In addition, the "condition" is a condition for determining the failure of the dialogue. The condition of "ID = 1" is that user utterance information having a similarity of 90% or more is continuously accepted. The process of calculating the similarity of user utterance information is, for example, the same as the process of calculating the similarity of sentences, and is a known technique, so detailed description thereof will be omitted. The condition of "ID = 2" is that there are 3 or more user utterance information having a similarity of 90% or more among the four user utterance information that are continuously accepted. The condition of "ID = 3" is that there are two or more automatic response information including the broken term among the four automatic response information that are continuously accepted. The condition of "ID = 4" is that the automatic response information including the information indicating that the conversation has not been established has been accepted. In addition, the broken term is assumed here, for example, to be "incomprehensible" or "incomprehensible". Further, it is assumed that the information indicating that the conversation is not established is, for example, a score "0" that can be included in the automatic response information. Further, for example, it is assumed that the default mode information of the mode storage unit 211 is "automatic mode".

Further, it is assumed that the storage unit 51 stores, for example, the priority information management table shown in FIG. Further, it is assumed that the storage unit 51 stores, for example, the selection condition management table shown in FIG.

Further, it is assumed that the learning information storage unit 61 constituting one dialogue engine 6 (for example, “e1”) stores the learning information management table shown in FIG. 7. Further, it is assumed that the learning information storage unit 61 stores the Foodtable shown in FIG. Foodtable stores one or more records having "ID", "cooking", and "how to make". Further, it is assumed that the learning information storage unit 61 has a food dictionary for storing the dish name and a crops dictionary for storing the crop name. The food dictionary stores a large number of dish names such as "Katsudon, Gyudon, Oyakodon, ...". In addition, a large number of crop names such as "rice, eggplant, carrot, ..." are stored in the crops dictionary.

In such a situation, the following three specific examples will be described. A specific example 1 is a case where the dialogue control device 2 automatically detects the failure of the dialogue and the mode information is automatically switched from the automatic mode to the manual mode. Further, in Specific Example 1, a process of automatically switching from the manual mode to the automatic mode will also be described. Further, a specific example 2 is a case where the administrator manually switches the mode information from the automatic mode to the manual mode. Further, in the specific example 2, a case where the administrator manually switches the mode information from the manual mode to the automatic mode will also be described. Further, the specific example 3 is a case where the dialogue engine 3 performs learning using the manual response information input by the administrator. In the three specific examples, for example, the dialogue engine 6 included in the dialogue system C is three.

(Specific example 1)
Now, it is assumed that the user has input the voice "Good morning" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 receives the utterance "good morning" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user utterance information "good morning". Next, the user transmission unit 13 transmits the user utterance information "good morning" to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 5 receives the user utterance information "good morning" from the user terminal 1.

Then, the processing unit 53 acquires the utterance information "1: Good morning" from the user utterance information "Good morning". That is, the processing unit 53 increments the stored utterance identifier "0" (assuming that it is the initial value of the utterance identifier) to generate a unique utterance identifier "1". Then, the processing unit 23 adds the utterance identifier "1" to the user utterance information to form the utterance information "1: Good morning".

Next, the determination unit 531 acquires the mode information "automatic mode" from the mode storage unit 511, and determines that the output destination of the utterance information is the dialogue engine 6.

Next, the selection unit 34 acquires the notification destination information of the three dialogue engines 6 from the table of FIG.

Next, the utterance information output unit 241 transmits the utterance information "1: Good morning" to the three dialogue engines 6.

Next, the utterance reception unit 32 of each of the three dialogue engines 6 receives the utterance information "1: Good morning" from the dialogue control device 5.

Next, for example, the response acquisition unit 33 of the dialogue engine "e1" acquires the automatic response information "good morning" paired with the "good morning" of the utterance information from the learning information management table of FIG. Here, it is preferable that the response acquisition unit 33 acquires the automatic response information "1: Good morning" to which the utterance identifier "1" is added.

Next, the response output unit 34 transmits the automatic response information “1: Good morning” acquired by the response acquisition unit 33 to the dialogue control device 5. The automatic response information to be sent may be "Good morning".

Next, the automatic response reception unit 251 of the dialogue control device 5 receives the automatic response information from the three dialogue engines 6. From the dialogue engine "e1", the automatic response reception unit 251 receives the automatic response information "1: Good morning".

Next, the selection unit 534 selects the automatic response information of the dialogue engine "e1" having a high priority by the selection algorithm described in the first embodiment.

Next, the processing unit 53 acquires the automatic information "Good morning" from the automatic response information of "e1". Next, the response transmission unit 261 transmits the automatic information "Good morning" to the user terminal 1. The processing unit 53 determines that the automatic information is not output to the management terminal 7.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information "Good morning". Then, the user output unit 15 outputs the automatic information "Good morning" by voice.

Next, it is assumed that the user voice-inputs "Is it open next Thursday?" To the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Is it open next Thursday?" From the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user utterance information "Is it open next Thursday?". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 2.

Next, the receiving unit 22 of the dialogue control device 5 receives the user utterance information "Is it open next Thursday?" From the user terminal 1.

Then, the processing unit 53 acquires the utterance information "2: Is it open next Thursday?" From the user utterance information. That is, the processing unit 23 increments the stored utterance identifier "1" to generate a unique utterance identifier "2". Then, the processing unit 23 adds the utterance identifier "2" to the user utterance information, and constitutes the utterance information "2: Is it open next Thursday?".

Next, the determination unit 531 determines the output destination of the utterance information from the mode information as the "dialogue engine 6".

Next, the utterance information output unit 241 transmits the utterance information "2: Is it open next Thursday?" To each of the three dialogue engines 3.

Next, the utterance reception unit 32 of each of the three dialogue engines 3 receives the utterance information "2: Is it open next Thursday?" From the dialogue control device 2.

Next, the response acquisition unit 33 of each of the three dialogue engines 3 acquires the automatic response information corresponding to the utterance information "Is it open next Thursday?".

In addition, in the response acquisition unit 33 of the dialogue engine "e1", "Is it open next Thursday?" Is open as the pattern of utterance information in the learning information management table in Fig. 8 "{timeExpress} *. Detects that it matches "? *". The pattern of the utterance information is the information of "ID = 3" in the learning information management table of FIG. Then, the response acquisition unit 33 acquires the automatic response information of "ID = 3". Then, the response acquisition unit 33 substitutes "next Thursday" for the {timeExpression} of the automatic response information, and starts from the value of "$ business day" "Sunday, Tuesday, Wednesday, Thursday, Friday, Saturday". It is determined that next Thursday is included in the business day (that is, the execution result of the automatic response information is "true"). The response acquisition unit 33 extracts "Thursday" indicating the day of the week from "Next Thursday" and determines that it is included in the value of "$ business day". Next, the response acquisition unit 33 of the dialogue engine "e1" acquires the pattern of the automatic response information "Yes. {TimeExpression} is open". Next, the response acquisition unit 33 substitutes "next Thursday" for the {timeExpression} of the automatic response information pattern, and acquires the automatic response information "next Thursday is open". Then, the response transmission unit 261 transmits the automatic response information "2: Open next Thursday" to the dialogue control device 5.

Next, the automatic response reception unit 551 of the dialogue control device 5 receives the automatic response information from each of the three engines 3.

Next, the selection unit 534 selects the automatic response information "2: next Thursday is open" from the three automatic response information by the above-mentioned processing.

Next, the processing unit 53 acquires the automatic information "Open next Thursday" from the automatic response information. Next, the determination unit 531 determines to adopt the dialogue engine 6 from the mode information "automatic mode". Then, the response transmission unit 561 transmits the automatic information "Open next Thursday" to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information "Open next Thursday". Then, the user output unit 15 outputs the automatic information "Open next Thursday" by voice.

As described above, it is assumed that the operation of the dialogue between the user and one of the three dialogue engines 6 is repeatedly performed. Then, it is assumed that the information including "incomprehensible" in all the automatic response information transmitted from the three dialogue engines 6 is transmitted to the dialogue control device 2 twice in succession. It is assumed that each of the three dialogue engines 6 configures and transmits automatic response information including "incomprehensible" when the automatic response information corresponding to the received utterance information cannot be acquired. Further, it is assumed that the user utterance information corresponding to the second automatic response information including "incomprehensible" was "Is it possible to reserve a seat next Thursday?".

Then, it is assumed that the automatic response reception unit 551 of the dialogue control device 5 has automatically selected the automatic response information including the second "incomprehensible". It is preferable that the processing unit 53 stores logs such as selected automatic response information, user utterance information, and manual response information in the storage unit 51. Alternatively, the processing unit 53 may store the automatic response information in the storage unit 51 when the automatic response information including the first "incomprehensible" is selected.

Next, the determination unit 532 determines that the automatic response information including two consecutive "incomprehensible" matches the failure condition of "ID = 3" in FIG.

Next, the mode changing unit 533 changes the mode information "automatic mode" of the mode storage unit 211 to "manual mode".

Next, the processing unit 53 determines that the change is to the manual mode. Then, the output unit 54 outputs the latest utterance information "12: Is it possible to reserve a seat next Thursday?" To the management terminal 7. In addition, "12" is an utterance identifier.

Next, the management information reception unit 71 of the management terminal 7 receives the utterance information "12: Is it possible to reserve a seat next Thursday?" From the dialogue control device 2.

Next, the administrator output unit 73 outputs the utterance content "Is it possible to reserve a seat next Thursday?" In the utterance information. An example of such an output is shown in FIG. In FIG. 20, the utterance content of the utterance information is displayed in 2001.

Next, it is assumed that the administrator presses the "Yes" button of 2002 in FIG. 20. Then, the management information reception unit 71 detects that the "Yes" button is pressed, and acquires the manual response information "Yes." Stored in association with the "Yes" button from the manual response information management table of FIG. To do. The management storage unit 70 manages manual response information corresponding to the buttons on the screen of FIG. 20. The manual response information management table is shown in FIG. In FIG. 21, records having "ID", "button", and "manual response information" are managed. Each attribute value of the "button" corresponds to the button of FIG. Next, the administrator output unit 73 transmits the manual response information “Yes.” To the dialogue control device 5.

Here, the administrator does not press the button in the screen of FIG. 20, or after pressing the "voice input" button of 2003 of FIG. 20, "Yes. Next Thursday, up to 4 people. You can make a reservation by voice input. In such a case, the management information reception unit 71 of the dialogue control device 2 receives the voice data "Yes. Up to 4 people can be reserved next Thursday.", Recognizes the voice data, and manually responds to the text. Get the information "Yes. Next Thursday, up to 4 people can be booked." Then, the administrator output unit 73 transmits the manual response information to the dialogue control device 5. The management information reception unit 71 may receive voice by default.

Next, the manual information reception unit 552 of the dialogue control device 5 receives the manual response information "Yes. Up to 4 people can be reserved next Thursday." Then, the processing unit 53 acquires the manual response information as manual information.

Then, since the mode information is "manual mode", the response transmission unit 561 transmits the manual information "Yes. Up to 4 people can be reserved next Thursday." To the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the manual information. Next, the user output unit 15 outputs the manual information "Yes. Next Thursday, up to 4 people can be reserved" by voice.

By the above processing, the failure of the dialogue is automatically detected, and the dialogue can be continued by automatically switching from the automatic mode to the manual mode.

Further, in the state of the manual mode, when the dialogue (reception of the manual response information) is performed more than the threshold value, the determination unit 532 determines that the mode information is changed from the "manual mode" to the "automatic mode". Then. Then, the mode change unit 533 changes the mode information of the mode storage unit 511 from the "manual mode" to the "automatic mode".

After that, the dialogue processing in the "automatic mode" as described above is continued. It is preferable that the user terminal 1 outputs automatic information or manual information by voice as the character speaks as shown in FIG.

(Specific example 2)
Now, it is assumed that the user voice-inputs "Tell me how to make beef bowl" to the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the utterance "Tell me how to make beef bowl" from the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user's utterance information "Tell me how to make beef bowl". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 5.

Next, the receiving unit 22 of the dialogue control device 5 receives the user utterance information "Tell me how to make beef bowl" from the user terminal 1.

Then, the processing unit 53 acquires the utterance information "1: Tell me how to make beef bowl" from the user's utterance information. The "1" in the utterance information is the utterance identifier.

Next, the determination unit 531 determines the output destination of the utterance information to the dialogue engine 6 and the management terminal 7. In this specific example, it is assumed that the output destination of the utterance information is always the dialogue engine 6 and the management terminal 7.

Next, the selection unit 534 selects one dialogue engine 6 from the three dialogue engines 6. Here, the selection unit 534 performs natural language processing on "Tell me how to make beef bowl" and acquires the genre "cooking". Then, the selection unit 534 selects the dialogue engine "e2" that is paired with the genre "cooking" with reference to FIG. In addition, the selection unit 534 performs morphological analysis of, for example, "teach me how to make beef bowl", acquires the independent words "gyudon", "how to make", and "teach", and machine learning (for example, deep layer) from the independent words. Acquire the genre "cooking" by learning, decision tree, SVM).

Next, the utterance information output unit 541 transmits the utterance information "1: Tell me how to make beef bowl" to the dialogue engine "e2" and the management terminal 7.

Then, the management terminal 7 receives the utterance information "1: Tell me how to make beef bowl" and outputs "Tell me how to make beef bowl" included in the utterance information.

Next, the utterance reception unit 32 of the dialogue engine "e2" receives the utterance information "1: Tell me how to make beef bowl" from the dialogue control device 5.

Next, the response acquisition unit 33 of the dialogue engine "e2" morphologically analyzes "tell me how to make beef bowl" that the utterance information has. Then, the response acquisition unit 33 searches the dictionary using the acquired morpheme "gyudon" as a key, and detects that "gyudon" is included in {food}. Next, the response acquisition unit 33 detects that "Tell me how to make beef bowl" matches the utterance information of "ID = 4" in the learning information management table of FIG. 7. Then, the response acquisition unit 33 acquires the automatic response information of "ID = 4", executes the automatic response information, and prepares "Gyudon" corresponding to the dish "Gyudon" from the Foodtable of FIG. Introducing ... "is acquired. Then, the response output unit 34 transmits the automatic response information "1: 4 servings of material ...." acquired by the response acquisition unit 33 to the dialogue control device 5.

Next, the reception unit 55 of the dialogue control device 5 receives the automatic response information. Then, the processing unit 53 acquires the automatic information "Introducing the material for 4 servings ..." from the answer information. Here, the automatic response information and the automatic information may be the same.

Next, the processing unit 53 determines that the automatic information is output to the management terminal 7. Then, the automatic information output unit 542 outputs the automatic information "Introducing the material for 4 servings ..." to the management terminal 7.

Next, the determination unit 531 determines that the dialogue engine 6 is adopted from the mode information "automatic mode". Then, the response transmission unit 561 transmits the automatic information "Introducing the material for four servings ..." from the dialogue engine "e2" to the user terminal 1.

Next, the user receiving unit 14 of the user terminal 1 receives the automatic information. Then, the user output unit 15 outputs the automatic information.

Next, it is assumed that the user voice-inputs "Which soup is recommended for powdered udon soup?" To the user terminal 1. Then, the user reception unit 11 of the user terminal 1 accepts the voice input of the user. Next, the user processing unit 12 voice-recognizes the user's voice input and acquires the user utterance information "Which soup is recommended for powdered udon soup?". Next, the user transmission unit 13 transmits the user utterance information to the dialogue control device 5.

Next, the receiving unit 22 of the dialogue control device 5 receives the user utterance information from the user terminal 1.

Then, the processing unit 53 acquires the utterance information "2: Which soup is recommended for the powdered udon soup?" From the user's utterance information.

Next, the determination unit 531 determines the output destination of the utterance information to the dialogue engine 6 and the management terminal 7. Then, the utterance information output unit 541 transmits the utterance information to the selected dialogue engine "e2" and the management terminal 7.

Next, the dialogue engine "e2" receives the utterance information, acquires the automatic response information corresponding to the utterance information, and transmits it to the dialogue control device 5. It should be noted that the automatic response information here includes, for example, information to the effect that "it cannot be understood".

Further, the management terminal 7 receives the utterance information and outputs "Which soup is recommended for the powdered udon soup?" Included in the utterance information.

Then, the administrator who sees this output determines that he / she cannot answer any of the dialogue engines 6, and inputs the mode information change instruction to the management terminal 7. Then, the administrator supposes that "A company's soup is recommended for powdered udon soup."

Next, the management instruction receiving unit 74 receives a mode change instruction (change instruction from the automatic mode to the manual mode) from the administrator. Then, the management output unit 75 outputs such a change instruction to the dialogue control device 5.

In addition, the management information reception unit 71 receives the voice data "Powdered udon soup is recommended from company A.", Recognizes the voice data, and changes it into a text. Next, the administrator output unit 73 transmits the text manual response information "2: The soup of company A is recommended for the powdered udon soup" to the dialogue control device 5.

Next, the change instruction receiving unit 553 of the dialogue control device 5 receives the change instruction from the management terminal 7. The mode change unit 533 changes the mode information of the mode storage unit 51 to the manual mode.

Further, the manual response reception unit 552 receives the manual response information from the management terminal 7. The processing unit 53 acquires the automatic information and the manual information from the received automatic response information and the manual response information.

Next, the determination unit 531 determines that the management terminal 7 is adopted from the mode information "manual mode".

Then, the response transmission unit 561 transmits the acquired manual information "Powdered udon soup is recommended from company A's soup" to the user terminal 1.

Next, the user terminal 1 receives and outputs the manual information "Powdered udon soup is recommended from company A."

Through the above processing, the administrator can appropriately intervene in the dialogue, and an effective and continuous dialogue can be performed by using a plurality of dialogue engines. In this specific example, after the administrator inputs an instruction to change the mode information in the "automatic mode", the dialogue can be performed in the "automatic mode".

(Specific example 3)
In Specific Example 2, after the manual response reception unit 552 of the dialogue control device 5 receives the manual response information "2: Powdered udon soup is recommended from company A" from the management terminal 7, the processing unit 53 receives the manual response information. It is determined that the manual response information is transmitted to each of the three dialogue engines 6.

Then, the manual information output unit 543 transmits such manual response information to each of the three dialogue engines 6. Here, the manual information output unit 543 may transmit the manual information to each of the three dialogue engines 6. Further, the manual information output unit 543 normally transmits the utterance information or the user utterance information to each of the three dialogue engines 6 in pairs with the manual response information or the manual information. It is preferable that the manual information output unit 543 does not transmit the utterance information or the user utterance information to the dialogue engine 6 to which the utterance information has already been transmitted.

Next, the learning reception unit 65 of the dialogue engine 6 receives the manual response information "2: The soup of company A is recommended for the powdered udon soup."

Next, the learning unit 66 configures information to be added to the learning information by using the manual response information received by the learning reception unit 65. That is, the learning unit 66 acquires the utterance information corresponding to the utterance identifier "2" of the manual response information "2: The soup of company A is recommended for the powdered udon soup." Then, the content of the utterance information corresponding to the utterance identifier "2" "Which soup is recommended for the powdered udon soup?" And the content of the manual response information "The soup of company A is recommended for the powdered udon soup." The set with is added to the learning information management table of FIG.

In addition, the learning unit 66 replaces the nomenclature or the nomenclature "powdered udon soup" and "soup" with variables in the content of the utterance information "Which soup is recommended for powdered udon soup?" } Is changed to the pattern of "Which {variable 2} is recommended?", And among "Powdered udon soup is recommended from company A.", Nomenclature or utterances "Powdered udon soup" "A" Replace "company" and "soup" with variables, change to the pattern "{variable 3} is recommended for {variable 4} {variable 5}", and change the set of the two patterns to the learning information management shown in Fig. 7. You may add it to the table.

That is, in this specific example, the learning information may be improved by using the manual response information input from the administrator, and the structure of the learning information does not matter.

Further, in this specific example, it is preferable that each dialogue engine 6 performs learning processing using the automatic response information output by the other dialogue engine 6.

As described above, according to the present embodiment, an appropriate dialogue can be continued by properly using the response from the dialogue engine 6 and the response from the administrator and outputting them to the user terminal.

Further, according to the present embodiment, it is possible to provide a dialogue engine 6 capable of learning a response from an administrator.

Further, according to the present embodiment, it is possible to provide a management terminal 7 capable of supporting the dialogue, for example, when the dialogue breaks down.

The software that realizes the dialogue control device 5 in this embodiment is the following program. That is, this program is a program in a dialogue control device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines, and uses a computer to display user utterance information corresponding to the user's utterance. The receiving unit received from the user terminal, the utterance information output unit capable of outputting the utterance information which is the information corresponding to the user utterance information to one or more dialogue engines, and the utterance information output unit according to the output of the utterance information. The user receives an automatic response reception unit that receives automatic response information from one or more dialogue engines, and one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit. It is a program for functioning as a response transmitter to be transmitted to a terminal.

Further, in the above program, the utterance information output unit can also output the utterance information to the management terminal, and the computer can output the utterance information from the management terminal in response to the output of the utterance information to the management terminal. Further functioning as a manual response receiving unit that receives manual response information that is a response, the response transmitting unit is information corresponding to the one automatic response information, one automatic information, or information corresponding to the manual response information. It is preferable that the program operates the computer to transmit the manual information to the user terminal.

Further, in the above program, when the computer further functions as a decision unit for deciding whether to adopt the dialogue engine or the management terminal, and the response transmission unit determines that the decision adopts the dialogue engine. Is a program that causes the computer to function by transmitting the above-mentioned automatic information to the user terminal and, in the case of the decision to adopt the management terminal, transmitting the manual information to the user terminal. Is suitable.

Further, in the above program, in order to change the mode information and the mode storage unit in which at least the mode information that can be taken is changed between the automatic mode that employs the dialogue engine and the manual mode that employs the management terminal. Further functions as a determination unit for determining whether or not the predetermined condition is satisfied, and a mode change unit for changing the mode information when the determination unit determines that the predetermined condition is satisfied. It is preferable that the determination unit is a program that causes the computer to function as a device for acquiring the mode information of the mode storage unit.

Further, in the above program, the computer has either an automatic mode that employs a dialogue engine or a manual mode that employs a management terminal, at least a mode storage unit that stores mode information that can be taken, and a mode change from the management terminal. The change instruction receiving unit that receives the change instruction, which is the instruction of the above, and the mode changing unit that changes the mode information of the mode storage unit in response to the change instruction are further functioned, and the determining unit is the mode of the mode storage unit. It is preferable that the program functions the computer as a decision to adopt the dialogue engine or the management terminal according to the information.

Further, in the above program, it is preferable that the computer further functions as an automatic information output unit that outputs the above-mentioned one automatic information to the management terminal, and as a program, the computer functions.

Further, in the above program, it is preferable that the computer further functions as a manual information output unit that outputs the manual information to the dialogue engine, and as a program, the computer functions.

(Embodiment 4)
In the present embodiment, the stand-alone dialogue device 8 will be described. Here, the dialogue device 8 usually has two or more dialogue engines 3. Further, the dialogue device 8 includes, for example, a user terminal 1 of the dialogue system A, a dialogue control device 2, and a dialogue engine 3. However, the dialogue device 8 here does not include the management terminal 7 and communicates with the management terminal 7.

FIG. 22 is a block diagram of the dialogue device 8 according to the present embodiment.

The dialogue device 8 includes a reception unit 81, an utterance information output unit 82, an automatic response reception unit 83, a manual response reception unit 84, a response output unit 85, and one or more dialogue engines 3.

The reception unit 81 receives the user utterance information corresponding to the utterance input by the user. The reception unit 81 may accept utterances and configure user utterance information from the utterances. Further, the reception here may be reception of user utterance information which is voice data, or reception of text user utterance information. Here, acceptance means accepting information input from an input device such as a keyboard, mouse, or touch panel, receiving information transmitted via a wired or wireless communication line, and recording an optical disk, magnetic disk, semiconductor memory, or the like. It is a concept including acceptance of information read from a medium.

The input means of the user utterance information may be any means such as a keyboard, a mouse, and a menu screen. The reception unit 81 can be realized by a device driver for input means such as a numeric keypad or a keyboard, control software for a menu screen, or the like.

The utterance information output unit 82 outputs utterance information related to user utterance information to at least one of two or more dialogue engines 3 or a screen. The output here may be the delivery of utterance information. The output to the screen may be the output to the screen of the management terminal 7.

The utterance information output unit 82 can usually be realized from an MPU, a memory, or the like. The processing procedure of the utterance information output unit 82 is usually realized by software, and the software is recorded in a recording medium such as ROM. However, it may be realized by hardware (dedicated circuit). The utterance information output unit 82 may be realized by a wireless or wired communication means.

The automatic response reception unit 83 receives automatic response information which is a response from one or more dialogue engines 3 in response to the output of the utterance information to one or more dialogue engines 3. The automatic response reception unit 83 has a function of a selection unit 231 that selects one automatic response information from a plurality of automatic response information, or a selection unit 431 that selects one dialogue engine from a plurality of dialogue engines.

The manual response reception unit 84 receives the manual response information, which is a response from the administrator, in response to the output of the utterance information to the screen. The reception here may be reception from the management terminal 7.

The manual response receiving unit 84 can be realized by, for example, a receiving means of a wireless or wired network.

The response output unit 85 outputs automatic information regarding the automatic response information or manual information regarding the manual response information. The output here is an output to the user, for example, an audio output or a display on a display.

The response output unit 85 may or may not include an output device such as a display or a speaker. The response output unit 85 can be realized by the driver software of the output device, the driver software of the output device, the output device, or the like.

Next, the operation of the dialogue device 8 will be described with reference to the flowchart of FIG. In the flowchart of FIG. 23, the same steps as those of the flowcharts of FIGS. 4 and 11 will be omitted.

(Step S2301) The reception unit 81 determines whether or not the user utterance information has been received. If the user utterance information is accepted, the process goes to step S402, and if the user utterance information is not accepted, the process goes to step S419. In this step, the reception unit 81 may accept utterances from the user and configure user utterance information from the utterances. Such processing is also the reception of user utterance information.

(Step S2302) The automatic response reception unit 83 determines whether or not to operate the dialogue engine 3. If the dialogue engine 3 is operated, the process goes to step S2303. If the dialogue engine 3 is not operated, the process goes to step S2304. The automatic response reception unit 53 determines that the dialogue engine 3 is operated, for example, when the mode information of the storage unit 51 is the automatic mode. Further, when operating the dialogue engine 3, the automatic response reception unit 83 may operate a plurality of dialogue engines 3 or may operate one selected dialogue engine 3. The process of selecting the dialogue engine 3 has been described above.

(Step S2303) The automatic response receiving unit 53 acquires the automatic response information which is the output of one dialogue engine 3. When the automatic response receiving unit 83 operates a plurality of dialogue engines 3, usually, one automatic response information is selected from two or more automatic response information. The process of selecting one automatic response information has been described above.

(Step S2304) The manual response reception unit 84 acquires the manual response information corresponding to the input of the administrator. For example, when the mode information is the manual mode, the manual response receiving unit 84 waits until the manual response information can be acquired.

(Step S2305) The response output unit 85 determines whether or not to output the automatic information to the screen. If it is output to the screen, it goes to step S2306, and if it is not output, it goes to step S2307. For example, a flag indicating whether or not to output automatic information to the screen is managed by the storage unit 51, and the response output unit 85 reads such a flag from the storage unit 80 and determines. Further, the response output unit 85 may always output the automatic information to the screen, or may not always output the automatic information to the screen. Further, although the processing of this step is performed by the response output unit 85, a determination means (not shown) may be used.

(Step S2306) The response output unit 85 outputs automatic information to the screen.

(Step S2307) A processing unit (not shown) determines whether or not to output manual information to the dialogue engine 3. If it is output to the dialogue engine 3, it goes to step S2308, and if it is not output, it goes to step S2309. For example, a flag indicating whether or not to output manual information to the dialogue engine 3 is managed by the storage unit 80, and the processing unit 23 reads such a flag from the storage unit 21 and determines. Further, the processing unit 23 may always output the manual information to the dialogue engine 3, or may not always output the manual information to the dialogue engine 3.

(Step S2308) A processing unit (not shown) outputs the acquired manual information to the dialogue engine 3.

(Step S2309) The response output unit 8 acquires information to be output to the user. Such information may be automatic or manual information. Further, the response output unit 85 acquires, for example, automatic information when the mode information is in the automatic mode, and acquires manual information when the mode information is in the manual mode.

(Step S2310) The response output unit 85 outputs the information acquired in step S2309.

(Step S2311) A determination unit (not shown) determines whether or not the condition for changing the mode information is satisfied.

(Step S2312) If the determination result in step S2311 is a determination result that the condition is satisfied, the process goes to step S2312, and if the determination result is that the condition is not satisfied, the process goes to step S2314.

(Step S2313) The mode changing unit (not shown) changes the mode information of the storage unit 80.

(Step S2314) A processing unit (not shown) determines whether or not the change in step S2313 is a change to the manual mode. If it is a change to the manual mode, it goes to step S2315, and if it is not a change to the manual mode, it goes to step S2316.

(Step S2315) The utterance information output unit 82 outputs the latest utterance information to the screen. Return to step S2302.

(Step S2316) The utterance information output unit 82 outputs the latest utterance information to each dialogue engine 3. Return to step S2302.

In the flowchart of FIG. 23, the process ends when the power is turned off or an interrupt for the end of the process occurs.

As described above, according to the present embodiment, an appropriate dialogue can be continued by properly using the response from one or two or more dialogue engines 3 and the response from the administrator and outputting them to the user.

In the present embodiment, the dialogue device 8 communicates with the management terminal 7. However, the dialogue device 8 may communicate with the user terminal 1. In such a case, the dialogue device 8 may include a management terminal 7.

Further, the software that realizes the dialogue device 8 in the present embodiment is the following program. That is, this program is a program that realizes a dialogue device that can receive automatic response information that is a response corresponding to a user's utterance from two or more dialogue engines, and uses a computer as information corresponding to the user's utterance information. An utterance information output unit that can output certain utterance information to one or more dialogue engines, an automatic response reception unit that receives automatic response information from each of the one or more dialogue engines in response to the output of the utterance information, and an automatic response reception unit. This is a program for functioning as a response output unit that outputs one automatic information corresponding to one of the one or more automatic response information received by the automatic response reception unit to the user terminal.

In the above program, the utterance information output unit outputs utterance information related to the user utterance information to at least one of one or more dialogue engines or screens, and responds to the output of the utterance information to the screen. The computer further includes a manual response receiving unit that receives manual response information that is a response from the administrator, and the response output unit outputs automatic information about the automatic response information or manual information about the manual response information. It is preferable that the program is to function.

Further, FIG. 24 shows the appearance of a computer that executes the program described in the present specification to realize the dialogue control device and the like of the various embodiments described above. The embodiments described above may be implemented in computer hardware and computer programs running on it. FIG. 24 is an overview view of the computer system 300, and FIG. 25 is a block diagram of the system 300.

In FIG. 24, the computer system 300 includes a computer 301 including a CD-ROM drive, a keyboard 302, a mouse 303, and a monitor 304.

In FIG. 25, in addition to the CD-ROM drive 3012, the computer 301 is connected to the MPU 3013, the bus 3014 connected to the CD-ROM drive 3012, the ROM 3015 for storing a program such as a bootup program, and the MPU 3013. It includes a RAM 3016 that is connected and for temporarily storing instructions of an application program and providing a temporary storage space, and a hard disk 3017 for storing an application program, a system program, and data. Although not shown here, the computer 301 may further include a network card that provides a connection to the LAN.

A program for causing the computer system 300 to execute a function such as the dialogue control device of the above-described embodiment may be stored in the CD-ROM 3101, inserted into the CD-ROM drive 3012, and further transferred to the hard disk 3017. Alternatively, the program may be transmitted to the computer 301 via a network (not shown) and stored on the hard disk 3017. The program is loaded into RAM 3016 at run time. The program may be loaded directly from the CD-ROM3101 or network.

The program does not necessarily include an operating system (OS) that causes the computer 301 to execute functions such as the dialogue control device of the above-described embodiment, or a third-party program or the like. The program need only include a portion of the instruction that calls the appropriate function (module) in a controlled manner to obtain the desired result. It is well known how the computer system 300 works, and detailed description thereof will be omitted.

In the above program, in the step of transmitting information and the step of receiving information, processing performed by hardware, for example, processing performed by a modem or interface card in the transmission step (performed only by hardware). Processing) is not included.

Further, the number of computers that execute the above program may be singular or plural. That is, centralized processing may be performed, or distributed processing may be performed.

Further, it goes without saying that in each of the above embodiments, the two or more communication means existing in one device may be physically realized by one medium.

Further, in each of the above-described embodiments, each process may be realized by centralized processing by a single device, or may be realized by distributed processing by a plurality of devices.

It goes without saying that the present invention is not limited to the above embodiments, and various modifications can be made, and these are also included in the scope of the present invention.

As described above, the dialogue control device according to the present invention has an effect that an appropriate dialogue can be continued by effectively using a plurality of dialogue engines, and is useful as a dialogue control device or the like.

1 User terminal 2, 4, 5, 6 Dialogue control device 3, 6 Dialogue engine 7 Management terminal 8 Dialogue device 11 User reception unit 12 User processing unit 13 User transmission unit 14 User reception unit 15 User output unit 21, 41, 51, 80 Storage unit 22 Reception unit 23, 43, 53 Processing unit 24, 54 Output unit 25, 55, 81 Reception unit 26, 56 Transmission unit 31, 61 Learning information storage unit 32, 62 Speech reception unit 33, 63 Response acquisition unit 34 , 64, 85 Response output unit 35, 65 Learning reception unit 36, 66 Learning unit 51, 211,511 Mode storage unit 53, 83, 251, 551 Automatic response reception unit 70 Management storage unit 71 Management information reception unit 72 Management processing unit 73 Administrator output unit 74 Management instruction reception unit 75 Management output unit 82, 241, 541 Speech information output unit 84, 552 Manual response reception unit 231, 431, 534 Selection unit 242, 542 Automatic information output unit 242 Automatic response information output unit 261 and 561 Response transmission unit 531 Decision unit 532 Judgment unit 533 Mode change unit 543 Manual information output unit 552 Manual information reception unit 553 Change instruction reception unit

Claims (19)

A receiver that receives user utterance information corresponding to the user's utterance from the user terminal,
An utterance information output unit capable of outputting utterance information, which is information corresponding to the user utterance information, to two or more dialogue engines, and an utterance information output unit.
An automatic response reception unit that receives automatic response information from each of the two or more dialogue engines according to the output of the utterance information, and an automatic response reception unit.
A selection unit that selects one of the two or more automatic response information received by the automatic response reception unit, and a selection unit.
It includes a response transmission unit that transmits one automatic information corresponding to the one automatic response information selected by the selection unit to the user terminal.
By repeating the processes of the receiving unit, the utterance information output unit, the automatic response receiving unit, the selecting unit, and the response transmitting unit two or more times, dialogue control for continuing the dialogue with the user is performed. Dialogue control device. The selection unit
The dialogue control device according to claim 1, wherein one automatic response information that matches a predetermined condition is selected from the two or more automatic response information. Further, a storage unit for storing priority information indicating the priority of the two or more dialogue engines is provided.
The selection unit
Claim 1 for selecting the automatic response information output from the dialogue engine having the highest priority based on the priority information, among the automatic response information satisfying a predetermined condition among the two or more automatic response information. Alternatively, the dialogue control device according to claim 2. A receiver that receives user utterance information corresponding to the user's utterance from the user terminal,
A selection unit that selects one dialogue engine from two or more dialogue engines using the user utterance information, and
An utterance information output unit that outputs utterance information, which is information corresponding to the user utterance information, to one dialogue engine selected by the selection unit.
An automatic response reception unit that receives automatic response information from the one dialogue engine in response to the output of the utterance information, and an automatic response reception unit.
It is provided with a response transmission unit that transmits one automatic information corresponding to the automatic response information received by the automatic response reception unit to the user terminal.
By repeating the processes of the receiving unit, the selection unit, the utterance information output unit, the utterance information output unit, and the response transmission unit two or more times, dialogue control for continuing the dialogue with the user is performed. Dialogue control device. The utterance information output unit
The utterance information can also be output to the management terminal,
A manual response receiving unit that receives manual response information, which is a response from the management terminal, is further provided in response to the output of the utterance information to the management terminal.
The response transmitter is
The item according to any one of claims 1 to 4, wherein the one automatic information which is the information corresponding to the one automatic response information or the manual information which is the information corresponding to the manual response information is transmitted to the user terminal. Dialogue control device. It also has a decision-making unit that decides whether to adopt a dialogue engine or a management terminal.
The response transmitter is
If the decision decides to adopt the dialogue engine, the one automatic information is transmitted to the user terminal, and if the decision decides to adopt the management terminal, the manual information is transmitted to the user terminal. 5. The dialogue control device according to claim 5. A mode storage unit that stores at least possible mode information, either an automatic mode that employs a dialogue engine or a manual mode that employs a management terminal.
A judgment unit that determines whether or not a predetermined condition for changing the mode information is satisfied, and
When the determination unit determines that a predetermined condition is satisfied, the mode change unit for changing the mode information is further provided.
The decision unit
The dialogue control device according to claim 6, wherein the mode information of the mode storage unit is acquired. A mode storage unit that stores at least possible mode information, either an automatic mode that employs a dialogue engine or a manual mode that employs a management terminal.
A change instruction reception unit that accepts change instructions that are mode change instructions from the management terminal,
Further provided with a mode change unit that changes the mode information of the mode storage unit in response to the change instruction.
The decision unit
The dialogue control device according to claim 6, wherein the dialogue control device determines whether to adopt a dialogue engine or a management terminal according to the mode information of the mode storage unit. The dialogue control device according to any one of claims 5 to 8, further comprising an automatic information output unit that outputs the above-mentioned one automatic information to the management terminal. The dialogue control device according to any one of claims 5 to 9, further comprising a manual information output unit that outputs the manual information to the dialogue engine. A learning information storage unit that stores learning information, which is information for acquiring automatic response information corresponding to utterance information, and
An utterance receiving unit that receives utterance information from the dialogue control device according to any one of claims 1 to 10.
A response acquisition unit that acquires automatic response information, which is a response corresponding to the utterance information received by the utterance reception unit, using the learning information.
A response output unit that outputs the automatic response information acquired by the response acquisition unit to the dialogue control device, and a response output unit.
A learning reception unit that accepts other automatic response information that is a response from another dialogue engine,
A dialogue engine including a learning unit that adds information to the learning information by using other automatic response information received by the learning reception unit. A reception unit that accepts user utterance information corresponding to the user's utterance,
An utterance information output unit capable of outputting utterance information, which is information corresponding to the user utterance information, to two or more dialogue engines, and an utterance information output unit.
An automatic response reception unit that receives automatic response information from each of the two or more dialogue engines according to the output of the utterance information, and an automatic response reception unit.
A selection unit that selects one of the two or more automatic response information received by the automatic response reception unit, and a selection unit.
It is provided with a response output unit that outputs one automatic information corresponding to the one automatic information corresponding to the one automatic response information selected by the selection unit.
A dialogue device that interacts with the user by repeating the processes of the reception unit, the utterance information output unit, the automatic response reception unit, the selection unit, and the response output unit two or more times. A reception unit that accepts user utterance information corresponding to the user's utterance,
Using the user utterance information, a selection unit for selecting one of the dialog engine from two or more dialog engine,
The information der Ru speech information corresponding to the user utterance information, the utterance information output unit that outputs one of the dialog engine, wherein the selecting unit selects,
An automatic response reception unit that receives automatic response information from the one dialogue engine in response to the output of the utterance information, and an automatic response reception unit.
It is provided with a response output unit that outputs one automatic information corresponding to the automatic response information received by the automatic response reception unit.
A dialogue device that interacts with the user by repeating the processes of the reception unit, the selection unit, the utterance information output unit, the automatic response reception unit, and the response output unit two or more times. It is a dialogue control method realized by a receiving unit, an utterance information output unit, an automatic response receiving unit, a selection unit, and a response transmitting unit.
A reception step in which the receiving unit receives user utterance information corresponding to the user's utterance from the user terminal, and
An utterance information output step in which the utterance information output unit can output utterance information, which is information corresponding to the user utterance information, to two or more dialogue engines.
An automatic response reception step in which the automatic response reception unit receives automatic response information from each of the two or more dialogue engines in response to the output of the utterance information.
A selection step in which the selection unit selects one of the two or more automatic response information received in the automatic response reception step.
The response transmission unit includes a response transmission step of transmitting one automatic information corresponding to the one automatic response information selected in the selection step to the user terminal.
Dialogue control for continuing dialogue with the user by repeating the reception step, the utterance information output step, the automatic response reception step, the selection step, and the response transmission step two or more times. Method. It is a dialogue control method realized by a receiving unit, a selection unit, an utterance information output unit, an automatic response receiving unit, and a response transmitting unit.
A reception step in which the receiving unit receives user utterance information corresponding to the user's utterance from the user terminal, and
The selection unit includes a selection step of using said user's utterance information, selects one interactive engine from two or more dialog engine,
The utterance information output unit, the information der Ru speech information corresponding to the user utterance information, the utterance information output step of outputting to an interactive engine in which the selecting unit selects,
An automatic response reception step in which the automatic response reception unit receives automatic response information from one dialogue engine selected in the selection step in response to the output of the utterance information.
The response transmission unit includes a response transmission step of transmitting one automatic information corresponding to the automatic response information received in the automatic response reception step to the user terminal.
Dialogue control for continuing dialogue with the user by repeating the reception step, the selection step, the utterance information output step, the automatic response reception step, and the response transmission step two or more times. Method. Computer,
A receiver that receives user utterance information corresponding to the user's utterance from the user terminal,
An utterance information output unit capable of outputting utterance information, which is information corresponding to the user utterance information, to two or more dialogue engines, and an utterance information output unit.
An automatic response reception unit that receives automatic response information from each of the two or more dialogue engines according to the output of the utterance information, and an automatic response reception unit.
A selection unit that selects one of the two or more automatic response information received by the automatic response reception unit, and a selection unit.
A program for functioning as a response transmission unit that transmits one automatic information corresponding to the one automatic response information selected by the selection unit to the user terminal.
By repeating the processes of the receiving unit, the utterance information output unit, the automatic response receiving unit, the selecting unit, and the response transmitting unit two or more times, dialogue control for continuing the dialogue with the user is performed. program. Computer,
A receiver that receives user utterance information corresponding to the user's utterance from the user terminal,
Using the user utterance information, a selection unit for selecting one of the dialog engine from two or more dialog engine,
The information der Ru speech information corresponding to the user utterance information, the utterance information output unit that outputs one of the dialog engine, wherein the selecting unit selects,
An automatic response reception unit that receives automatic response information from the one dialogue engine in response to the output of the utterance information, and an automatic response reception unit.
It functions as a response transmission unit that transmits one automatic information corresponding to the automatic response information received by the automatic response reception unit to the user terminal.
By repeating the processes of the receiving unit, the selecting unit, the utterance information output unit, the automatic response receiving unit, and the response transmitting unit two or more times, dialogue control for continuing the dialogue with the user is performed. program. Computer,
A reception unit that accepts user utterance information corresponding to the user's utterance,
An utterance information output unit capable of outputting utterance information, which is information corresponding to the user utterance information, to two or more dialogue engines, and an utterance information output unit.
An automatic response reception unit that receives automatic response information from each of the two or more dialogue engines according to the output of the utterance information, and an automatic response reception unit.
A selection unit that selects one of the two or more automatic response information received by the automatic response reception unit, and a selection unit.
It is made to function as a response output unit that outputs one automatic information corresponding to one automatic information corresponding to the one automatic response information selected by the selection unit.

A program that interacts with the user by repeating the processes of the reception unit, the utterance information output unit, the automatic response reception unit, the selection unit, and the response output unit two or more times. Computer,
A reception unit that accepts user utterance information corresponding to the user's utterance,
Using the user utterance information, a selection unit for selecting one of the dialog engine from two or more dialog engine,
The information der Ru speech information corresponding to the user utterance information, the utterance information output unit that outputs one of the dialog engine, wherein the selecting unit selects,
An automatic response reception unit that receives automatic response information from the one dialogue engine in response to the output of the utterance information, and an automatic response reception unit.
It functions as a response output unit that outputs one automatic information corresponding to the automatic response information received by the automatic response reception unit.
A program that interacts with the user by repeating the processes of the reception unit, the selection unit, the utterance information output unit, the automatic response reception unit, and the response output unit two or more times.

Images