JP7377668B2 - Control device, control method and computer program - Google Patents

Description

The present invention relates to technology used in customer service operations such as contact centers.

In recent years, multiple issues have been pointed out in contact center customer service operations that require manual intervention. For example, it has become difficult to secure human resources due to the decline in the working population and the high mental burden of dealing with complaints. Furthermore, due to conditions such as the high mental burden mentioned above, the turnover rate is extremely high and training costs are high.

To address these issues, technologies for customer service operations using AI have been proposed, as disclosed in Patent Document 1 and Patent Document 2, for example.

Japanese Patent Application Publication No. 2019-144400 JP2018-101941A

However, automation technology is not yet perfect, and the accuracy of processing such as recognition and interpretation may not be sufficient.
In view of the above circumstances, an object of the present invention is to provide a technology that allows more accurate processing to be executed in customer service operations.

One aspect of the present invention is to determine a response sentence to be presented to the customer based on the content of the customer's utterance, determine the business content in response to the customer's request, and cause the business execution device to execute the business content. The control unit is a control device that selects a device that will be the main body for executing processing for analyzing the utterance content from among a plurality of devices capable of performing the same type of processing.

One aspect of the present invention is the above-mentioned control device, wherein the content of the utterance is expressed as a voice, and the control unit is configured to cause a voice recognition device that executes a voice recognition process to perform voice recognition of the voice indicating the content of the utterance. The control unit requests processing, determines the response sentence based on the recognition result obtained from the voice recognition device, determines the business content, and when a predetermined condition is satisfied, the control unit From among the recognition devices, a device that is the main body for executing the voice recognition process is selected as a device that meets the conditions.

One aspect of the present invention is the above control device, wherein the control unit requests a speech synthesis device that executes speech synthesis processing to generate speech representing the response sentence, and obtains a speech from the speech synthesis device. The controller responds to the customer using the voice generated by the user, and if a predetermined condition is satisfied, the control unit selects a device that will be the main body for executing the voice synthesis process from among the plurality of voice synthesis devices. Select the device according to the above conditions.

One aspect of the present invention includes a first step of determining a response sentence to be presented to the customer based on the content of the customer's utterance, a second step of determining the business content in response to the customer's request, and a second step of determining the business content according to the customer's request. The third step is to instruct the business execution device to execute the content, and the device that is the main body for executing any of the processes in the first step, second step, or third step is divided into two or more devices that can perform the same type of processing. This control method includes a fourth step of selecting from among the devices.

One aspect of the present invention is a computer program for causing a computer to function as the above control device.

According to the present invention, it is possible to perform more accurate processing in customer service work.

1 is a schematic block diagram showing the system configuration of a customer service business system according to the present invention. FIG. 3 is a diagram showing a specific example of functional blocks of a flow control device. FIG. 3 is a diagram showing a specific example of a flow information table. FIG. 3 is a diagram showing a specific example of a rule information table. 2 is a sequence chart showing a specific example of processing executed by the customer service business system when a dialogue using voice is performed. 2 is a sequence chart showing a specific example of processing executed by the customer service business system when a dialogue using voice is performed.

Hereinafter, specific configuration examples of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic block diagram showing the system configuration of a customer service business system 100 of the present invention. The customer service business system 100 includes a response control device 10, a flow control device 20, a plurality of voice recognition devices 30 (30-1, 30-2), a plurality of intention interpretation devices 40 (40-1, 40-2), and a plurality of The speech synthesis device 50 (50-1, 50-2), a business execution device 60, a business system 70, and an operator terminal 80 are provided. In this way, regarding the speech recognition device 30, the intention interpretation device 40, and the speech synthesis device 50, a plurality of devices that perform the same type of processing are provided. As specific examples of devices that perform the same type of processing, two devices are shown in FIG. 1, but three or more devices may be provided. Each of the processes performed by the speech recognition device 30, the intention interpretation device 40, and the speech synthesis device 50 is a process of analyzing the content of the utterance. Each device of the customer service business system 100 is connected via a network 300 to enable data communication. The customer service system 100 performs customer service using voice dialogue or text dialogue for users (customers) of user terminals 200 connected via the network 300 or telephone network 400.

The user terminal 200 is communicably connected to the customer service business system 100. The user terminal 200 may be connected to the customer service system 100 for data communication via the network 300, or may be connected to the customer service system 100 for voice communication and data communication via the telephone network 400. May be connected. The user terminal 200 performs voice communication or text communication with the customer service business system 100 when operated by a user who is a customer. The user terminal 200 is an information processing device such as a smartphone, a telephone, a personal computer, or a smart speaker.

Next, each device of the customer service business system 100 will be explained.
The response control device 10 realizes a function as a communication interface between the user terminal 200 and the customer service business system 100. The response control device 10 manages a communication session with, for example, a user terminal 200. For example, when the user terminal 200 requests text dialogue communication, the response control device 10 outputs information indicating the characters transmitted from the user terminal 200 to the flow control device 20. For example, when the user terminal 200 requests voice dialogue communication, the response control device 10 outputs information indicating the voice transmitted from the user terminal 200 to the flow control device 20. When the response control device 10 receives the response information from the flow control device 20, it responds to the user terminal 200 based on the response information. For example, when the user terminal 200 requests text dialogue communication, the response control device 10 transmits a character string indicating the content of the response to the user terminal 200. For example, when the user terminal 200 requests voice dialogue communication, the response control device 10 responds with voice by transmitting an audio signal to the user terminal 200. Such functions of the response control device 10 may be realized using, for example, IVR (Interactive Voice Response).

The flow control device 20 performs corresponding processing according to the content of the user's utterances transmitted from the user terminal 200. FIG. 2 is a diagram showing a specific example of functional blocks of the flow control device 20. As shown in FIG. The flow control device 20 is configured using an information processing device such as a personal computer or a server device. The flow control device 20 includes a communication section 21 , a call history storage section 22 , a flow information storage section 23 , a rule information storage section 24 , and a control section 25 .

The communication unit 21 is configured using a communication interface. The communication unit 21 performs data communication with other devices via the network 300.

The call history storage unit 22 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The call history storage unit 22 stores call history information. The call history information includes call start time, call end time, user identification information, and call content information. The call start time is the time when the call started. In the case of text dialogue communication, this refers to the time when the text dialogue started. The call end time is the time when the call ended. In the case of text dialogue communication, this refers to the time when the text dialogue ended. The user identification information is information indicating the user who has interacted with the user. The user identification information may be, for example, identification information (user ID) assigned to the user in advance, a caller number (telephone number) assigned to the user terminal 200, or other information. It may be. The call content information includes the content of the customer's utterance and the content of the utterance to the customer by the customer service system 100 during the call between the customer and the customer service system 100. For example, in the case of a voice dialogue, the voice may be recorded data. For example, in the case of a character dialogue, the character string may be recorded data. When a voice dialogue is performed, both voice recording data and character string data indicating the result of voice recognition may be recorded.

The flow information storage unit 23 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The flow information storage unit 23 stores flow information. FIG. 3 is a diagram showing a specific example of the flow information table. The flow information table is flow information configured in a table format. The flow information table stores a plurality of flow information records 231. The flow information record 231 has a message ID and business content information. The message ID is identification information indicating the message to be notified to the customer. The business content information indicates the business content to be executed when the customer is notified of the message with the message ID recorded in the same flow information record 231. For example, if the message ID indicates a message indicating that a request has been received (for example, a message that says "We have accepted your request. Thank you."), the job content information may include the request content that has been accumulated in the previous voice. Information indicating that order processing is to be performed based on is defined. The business content information may be defined as information indicating a scenario of RPA (Robotic Process Automation), for example.

The rule information storage unit 24 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The rule information storage unit 24 stores rule information. FIG. 4 is a diagram showing a specific example of the rule information table. The rule information table is rule information configured in a table format. The rule information table stores a plurality of rule information records 241. The rule information record 241 includes conditions, execution timing, and device ID. The conditions indicate the conditions regarding the processing being performed with the customer. The execution timing indicates the timing at which the process based on the device ID is executed when the condition is met. The device ID is identification information that indicates the device for which the processing result is adopted. The device ID is, for example, identification information that uniquely indicates a device included in the customer service business system 100. The device ID includes, for example, identification information indicating the first speech recognition device 30-1, identification information indicating the second speech recognition device 30-2, identification information indicating the first intention interpretation device 40-1, and second intention interpretation device 40. -2, identification information indicating the first speech synthesis device 50-1, identification information indicating the second speech synthesis device 50-2, etc.

When the conditions indicated in the rule information record 241 are satisfied, the process indicated by the execution timing is executed by the device indicated by the device ID. For example, when message ID "A1" is output to a customer, control is performed such that the next process is executed by device ID "20d1". As in the case where the execution timing is "one time", the execution entity of the process is limited to the device indicated by the device ID for the process at the timing that indicates the process after the condition has been found to be satisfied. may be done.

For example, if the character string "name" is included in the voice response given by the customer, the voice will be processed by each device that can process it, and then the voice will be processed using the device ID "30a2". The execution result is adopted. If the execution timing is "current" and the process must be executed once to determine that the condition has been met, multiple (or all) execution entities may be responsible for executing the process. A device ID may be selected, and then a condition is determined, and only the execution result of the device ID corresponding to the satisfied condition may be adopted.

The control unit 25 is configured using a processor such as a CPU (Central Processing Unit) and a memory. The control unit 25 functions as a session control unit 251, a history control unit 252, a speech recognition control unit 253, an intention interpretation control unit 254, a speech synthesis control unit 255, and a business control unit 256 by executing a predetermined program.

The session control unit 251 controls the communication session between the user terminal 200 and the customer service business system 100. The session control unit 251 determines, for example, whether an escalation condition is satisfied. The escalation condition is a condition indicating that it may be difficult to maintain the customer response that is currently being performed. For example, the escalation condition may be that the voice received from the user terminal 200 includes a predetermined character string indicating that the customer may be offended. Escalation is the transfer of a customer interaction to another customer handler or other customer function. In this embodiment, as a specific example of escalation, customer service is transferred to a human operator. If the task control unit 256 determines that escalation is to be performed, the session control unit 251 changes the connection destination of the voice dialogue from the flow control device 20 to the operator terminal 80.

The history control unit 252 generates call history information regarding calls made between the user terminal 200 and the customer service business system 100. The history control unit 252 records the generated call history information in the call history storage unit 22.

The voice recognition control unit 253 requests the voice recognition device 30 to perform voice recognition processing on the voice transmitted from the user terminal 200 when a voice dialogue is taking place between the customer service business system 100 and the user terminal 200. do. Upon receiving the voice recognition result from the voice recognition device 30, the voice recognition control unit 253 outputs the voice recognition result to the intention interpretation control unit 254. When the condition stored in the rule information storage section 24 is satisfied and the device indicated by the device ID associated with the condition is a voice recognition device, the voice recognition control section 253 performs the following operation. , at the timing indicated by the execution timing, selects the speech recognition device 30 (executor) indicated by the device ID, and obtains the execution result of the speech recognition process by that device.

For example, in association with the condition of outputting a message ID indicating a predetermined message for which the customer's name is likely to be obtained as an answer, the device ID of the voice recognition device 30 with relatively high name recognition accuracy is associated. It may be. The predetermined message for which the customer's name is likely to be obtained as an answer is, for example, a message asking for the customer's name (for example, a message saying "Please tell me your name.").

For example, even if the device ID of the voice recognition device 30 with relatively high name recognition accuracy is associated with the condition that an answer that is likely to include the customer's name is obtained, good. An answer that is likely to include the customer's name is, for example, an answer that includes the character string "name."

For example, in association with the condition of outputting a message ID indicating a predetermined message for which the customer's address is likely to be obtained as an answer, the device ID of the voice recognition device 30, which has relatively high address recognition accuracy, is associated. It may be. The predetermined message for which the customer's address is likely to be obtained as an answer is, for example, a message asking for the address (for example, a message saying "Please tell me your address.").

For example, even if the device ID of the voice recognition device 30 with relatively high address recognition accuracy is associated with the condition that an answer that is likely to include the customer's address is obtained, good. An answer that is likely to include the customer's address is, for example, an answer that includes either the character string "address" or the character string of the name of one of the 47 prefectures.

The intention interpretation control unit 254 performs an intention interpretation process on a character string obtained as a result of voice recognition processing when a voice dialogue is performed between the customer service business system 100 and the user terminal 200. Request for 40. When the intention interpretation control unit 254 receives the intention interpretation result from the intention interpretation device 40, it outputs the character string of the intention interpretation result to the session control unit 251. When the condition stored in the rule information storage section 24 is satisfied and the device indicated by the device ID associated with the condition is an intention interpretation device, the intention interpretation control section 254 , at the timing indicated by the execution timing, selects the intention interpretation device 40 (execution subject) indicated by the device ID, and obtains the execution result of the intention interpretation process by that device.

For example, the device ID of the intention interpretation device 40 that has relatively high accuracy in interpreting the intention of the question may be associated with a condition that there is a high possibility that a voice indicating a question will be obtained from the customer. A situation in which there is a high possibility that a voice indicating a question will be obtained is, for example, a case where the customer selects an option indicating a question from among options indicating the purpose of the inquiry.

For example, the device ID of the intention interpretation device 40 that has relatively high precision in interpreting the intention of a complaint may be associated with the condition that there is a high possibility that a voice indicating a complaint will be obtained from the customer. A situation in which there is a high possibility that a voice indicating a complaint will be obtained is, for example, a case where the customer selects an option indicating a complaint from among options indicating the purpose of the inquiry.

When the data (response text information) obtained from the intention interpretation device 40 includes a person's name, the intention interpretation control unit 254 combines the person's name obtained in advance and the person's intention interpretation result. Compare with name. As a result of the comparison, if the two are the same, the intention interpretation control unit 254 proceeds with the process using the name of the person resulting from the intention interpretation. On the other hand, even if they are not the same as a result of the comparison, if the predetermined condition is satisfied that there is a high possibility that they are the same, the intention interpretation control unit 254 processes the person's name obtained in advance as the correct name. proceed. The predetermined condition that is likely to be the same may be, for example, that only one character string among the character strings of names is different. The predetermined condition that there is a high possibility of being the same may be, for example, that only one character string among the character strings of the name is different, and the vowels of the different character strings are the same. . The predetermined condition that there is a high possibility that they are the same means, for example, that only one character string in the character strings of a name is different, and that vowels in different character strings generally cause recognition errors. It may be an easy predetermined relationship (for example, "a" and "e").

The person's name obtained in advance is, for example, a name stored in advance in association with a user ID obtained by the customer by voice or operation of the user terminal 200. If the user ID is obtained by voice or operation of the user terminal 200 before the above processing by the intention interpretation control unit 254, the customer database registered in the own device or another device (for example, the business system 70) The name of the customer associated with may be acquired.

When receiving a speech synthesis instruction from the session control section 251, the speech synthesis control section 255 requests the speech synthesis device 50 to perform speech synthesis processing on the instructed character string. When the speech synthesis control section 255 receives the speech signal from the speech synthesis device 50, it outputs the speech signal to the session control section 251. The speech synthesis control section 255 controls the speech synthesis device 50 when the condition stored in the rule information storage section 24 is satisfied and the device indicated by the device ID associated with the condition is the speech synthesis device 50. selects the speech synthesis device 50 (executor) indicated by the device ID at the timing indicated by the execution timing, and obtains the execution result of the speech synthesis process by that device.

For example, the device ID of the speech synthesis device 50 that performs normal speech synthesis may be associated with a condition that there is a high possibility that a voice indicating a question will be obtained from the customer. A situation in which there is a high possibility that a voice indicating a question will be obtained is, for example, a case where the customer selects an option indicating a question from among options indicating the purpose of the inquiry.

For example, even if the device ID of the speech synthesis device 50 that satisfies a predetermined condition suitable for complaints is associated with the condition that there is a high possibility that voice indicating a complaint will be obtained from the customer. good. The speech synthesis device 50 that satisfies the predetermined conditions suitable for responding to a complaint may be, for example, a speech synthesis device 50 created for the purpose of outputting an apology speech, or a speech synthesis device 50 that is created for the purpose of outputting an apology speech, The speech synthesis device 50 may be capable of outputting a relatively low voice with a relatively slow reading speed. A situation in which there is a high possibility that a voice indicating a complaint will be obtained is, for example, a case where the customer selects an option indicating a complaint from among options indicating the purpose of the inquiry.

When the business control unit 256 receives a business control instruction from the session control unit 251, the business control unit 256 determines the content of the business to be executed based on the message ID accompanying the instruction. The determination of the work content is performed based on the flow information record stored in the flow information storage unit 23, for example. If there is a task to be executed, the task control unit 256 executes the task. For example, the business control unit 256 instructs the business execution device 60 to execute the business in a format that can be processed by the business execution device 60. When multiple types of business execution devices 60 are provided, the formats of the various business execution devices 60 may be stored in advance in the flow information storage unit 23. The task control unit 256 generates task instruction information in a format suitable for the task execution device 60 to which instructions are to be given. The task execution device 60 to be instructed may be determined based on, for example, the destination telephone number of the user terminal 200 or a subsequently selected push number.

Returning to FIG. 1, the description of each device of the customer service business system 100 will be continued. The speech recognition device 30 is configured using, for example, an information processing device such as a personal computer or a server device. The speech recognition device 30 is a device that provides speech recognition services to other devices. The speech recognition device 30 executes the speech recognition process upon receiving the instruction to perform the speech recognition process. The speech recognition device 30 returns information on a character string obtained as a recognition result.

The intention interpretation device 40 is configured using, for example, an information processing device such as a personal computer or a server device. The intention interpretation device 40 has learning results regarding natural language analysis and customer interaction, and executes intention interpretation processing using the learning results. Such learning results can be obtained, for example, by using machine learning techniques such as SVM (Support Vector Machine) and deep learning. The intention interpretation device 40 may be implemented using AI (Artificial Intelligence). The learning results may be held as multiple types of scenarios. For example, learning results regarding dialogue regarding ordering, learning results regarding dialogue regarding reservation confirmation, and learning results regarding dialogue regarding sales may be held as different learning results. Which learning result to use may be determined by the flow control device 20 based on, for example, the destination telephone number of the user terminal 200 or a subsequently selected push number. The intention interpretation device 40 determines the content of the voice dialogue by analyzing the recognition results of the customer's voice based on the learning results. If the intention interpretation device 40 cannot determine the content of the voice dialogue, it may transmit a response sentence for asking the question again. When the intention interpretation device 40 does not have all the information that is predetermined to be acquired, it may generate a response sentence for asking an additional question to find out the missing information. The intention interpretation device 40 determines a response sentence to be provided to the customer based on the content of the determined voice dialogue. The intention interpretation device 40 returns information indicating the determined response sentence to the flow control device 20.

The speech synthesis device 50 is configured using, for example, an information processing device such as a personal computer or a server device. The speech synthesis device 50 is a device that provides speech synthesis services to other devices. When the speech synthesis device 50 receives an instruction for speech synthesis processing, it generates an audio signal representing the instructed character string. The speech synthesis device 50 returns a speech signal (response message speech) obtained as a result of the speech synthesis processing to the flow control device 20.

The business execution device 60 is configured using, for example, an information processing device such as a personal computer or a server device. The business execution device 60 may be implemented as an RPA device, for example. When the task execution device 60 receives the task instruction information from the flow control device 20, it performs an input operation on the task system 70 based on the content of the received task instruction information. Various processes are executed in the existing business system 70 through the processing of the business execution device 60. For example, in the case of ordering a product, information necessary for the ordering process is input to the business system 70, and the process up to order execution is executed by the business execution device 60. The information necessary for the ordering process includes, for example, product identification information to be ordered (eg, product name, model number, ID), the number of items ordered, order source information (customer information), and the like.

The business system 70 is a system introduced by a corporation such as a company or an individual in order to carry out business. The business system 70 is a system used to provide services to customers who are users of the user terminal 200. Specific examples of the business system 70 include an ordering system for processing product orders, a reservation system for making reservations for hotels and stores, and a complaint response system for receiving complaints.

The operator terminal 80 is a terminal device used by an operator who deals with customers. The operator terminal 80 includes a user interface for inputting/outputting characters and inputting/outputting audio. When escalation is executed by the flow control device 20, a communication path is formed between the user terminal 200 and the operator terminal 80. Thereafter, the operator communicates with the customer using the user terminal 200 using the operator terminal 80. Upon receiving a business request from a customer, the operator instructs the business execution device 60 to execute the business by operating the operator terminal 80.

5 and 6 are sequence charts showing specific examples of processing executed by the customer service business system 100 when a dialogue using voice is performed. First, a communication path is formed for voice interaction between the user terminal 200 used by the customer and the customer service business system 100. When the customer speaks on the user terminal 200, the user terminal 200 converts the voice into an electrical signal and transmits it to the response control device 10 (step S100). The response control device 10 receives the audio electrical signal and records the received audio (step S101). The response control device 10 transmits an audio electrical signal to the flow control device 20 (step S102).

When the session control unit 251 of the flow control device 20 receives the audio electrical signal, it outputs the audio electrical signal to the history control unit 252 and the speech recognition control unit 253. The history control unit 252 temporarily records the content of the received audio. The voice recognition control unit 253 transmits the received voice signal to the voice recognition device 30 (step S103). At this time, if the conditions registered in the rule information storage unit 24 are satisfied, the voice recognition control unit 253 selects the voice recognition device 30 to which the voice signal is to be transmitted, depending on the timing. The speech recognition device 30 executes speech recognition processing (step S104). The speech recognition device 30 transmits the speech recognition result indicating the character string obtained by the speech recognition process to the flow control device 20 (step S105).

Upon receiving the voice recognition result, the voice recognition control unit 253 of the flow control device 20 outputs the voice recognition result to the intention interpretation control unit 254. The intention interpretation control unit 254 requests the intention interpretation device 40 to determine a response sentence based on the voice recognition result by transmitting the received voice recognition result (step S106). At this time, if the conditions registered in the rule information storage unit 24 are satisfied, the intention interpretation control unit 254 selects the intention interpretation device 40 to which response sentence determination is requested, depending on the timing. When the intention interpretation device 40 receives the voice recognition result, it determines the content of the voice dialogue by analyzing the voice recognition result, and determines a response sentence (step S107). For example, the intention interpretation device 40 estimates the question sentence closest to the utterance content by performing intention interpretation processing based on the voice recognition result. Then, the intention interpretation device 40 determines an appropriate response sentence to the estimated question sentence from among the FAQ list registered in advance in the database. The intention interpretation device 40 transmits information indicating the determined response sentence to the flow control device 20 (step S108).

When the intention interpretation control unit 254 of the flow control device 20 receives the information indicating the response sentence, it requests the speech synthesis device 50 to synthesize a voice representing the received response sentence (step S109). At this time, if the response sentence includes a person's name, the intention interpretation control unit 254 determines whether or not they are the same, and whether a predetermined condition that there is a high possibility that they are the same is satisfied. Furthermore, if the conditions registered in the rule information storage section 24 are satisfied, the speech synthesis control section 255 selects the speech synthesis device 50 to which the speech synthesis process is requested, depending on the timing. The speech synthesis device 50 generates speech data of a response message by performing speech synthesis processing according to the requested response sentence (step S110). The voice synthesis device 50 transmits the generated voice data of the response message to the flow control device 20 (step S111).

When the session control unit 251 of the flow control device 20 receives the voice data of the response message, it outputs the received voice data to the history control unit 252. The history control unit 252 temporarily records the audio content of the received response message. Note that the content of the audio temporarily recorded by the history control unit 252 is recorded in the call history storage unit 22 as call history information by the history control unit 252 after the communication path with the user terminal 200 is disconnected. Ru. The session control unit 251 transmits the voice of the response message to the response control device 10 (step S112). When the response control device 10 receives the voice from the flow control device 20, it records the received voice (step S113). The response control device 10 transmits the recorded voice to the user terminal 200 (step S114).

The session control unit 251 determines whether the escalation condition is satisfied at a predetermined timing (step S115). For example, the condition may be that the voice received from the user terminal 200 includes a predetermined character string indicating that the customer may be offended. If the escalation conditions are not met (step S115-NO), escalation is not performed. On the other hand, if the escalation conditions are satisfied (step S115-YES), the session control unit 251 instructs the response control device 10 to execute escalation. In response to this instruction, the response control device 10 executes escalation by changing the connection destination of the user terminal 200 to the operator terminal 80 (step S116). At this time, the session control unit 251 may transmit information regarding the call that is the subject of escalation to the operator terminal 80.

The work control unit 256 of the flow control device 20 determines the content of the work to be executed based on the response text information notified from the intention interpretation device 40 (step S201). For example, when the response text information includes a message ID indicating the content and type of the response message, the business control unit 256 can refer to the business content information recorded in the flow information storage unit 23 in association with the message ID. The business content may also be determined. If there is a task to be executed, the task control unit 256 generates task instruction information (step S202). The task instruction information is generated in a format that can be processed by the task execution device 60 to which the task is instructed. The task control unit 256 transmits the generated task instruction information to the task execution device 60 (step S203). When the task execution device 60 receives the task instruction information, it issues a task instruction to the business system 70 in accordance with the received task instruction information (step S204). The business system 70 executes the business according to the instructed content (step S205).

5 and 6 have explained the flow of processing assuming voice dialogue, but when text dialogue is performed, steps S103 to S105 and steps S109 to 111 are omitted, and steps S101 and S113 are omitted. In this process, the recording is performed as a character string rather than as a voice.

In the customer service business system 100 configured in this way, an optimal device is selected for executing a process according to the conditions registered in the rule information storage unit 24, and the processing results by the selected device are used for subsequent processing. Used in processing. Therefore, it becomes possible to perform more accurate processing in customer service operations.

In addition, in the customer service business system 100 configured in this way, even if the person's name obtained in advance and the person's name as a result of intention interpretation are different, the process will be processed if predetermined conditions are met. will be carried out. Therefore, it is possible to prevent subsequent processing from being stopped (for example, from being marked as an error) due to a processing error in speech recognition processing or intention interpretation processing.

Further, in the customer service business system 100, a response sentence is automatically determined based on experiential knowledge obtained in advance through machine learning in response to a voice or written statement uttered by a customer. If the conversation is a voice conversation, it is converted into voice and sent to the customer, and if it is a text conversation, it is sent to the customer as text. Therefore, it becomes possible to perform some of the customer service tasks that were previously performed manually without human intervention. As a result, it is possible to further reduce the manpower required for customer service work.

Further, the content of the so-called subsequent work is automatically determined based on the dialogue with the customer. Then, business instructions are given to business execution devices such as RPA devices in a format that can be interpreted by each device. Each business execution device executes a business in the business system according to instructions. Therefore, it becomes possible to automatically execute not only responses to customers but also subsequent operations. As a result, it is possible to further reduce the manpower required for customer service operations.

Further, business systems and business execution devices that are already available in companies and individuals that require customer business can be used as they are. Therefore, it is possible to reduce the cost and time necessary for introducing the customer service business system 100.

(Modified example)
The system configuration of the customer service business system 100 does not need to be limited to that shown in FIG. Specifically, the details are as follows.
Some of the devices included in the customer service business system 100 may be communicably connected to other devices via a network different from the network 300. For example, the business system 70 may be communicably connected to other devices via a network using a dedicated line.
Among the devices included in the customer service business system 100, some of the devices may be configured integrally with other devices. For example, the speech recognition device 30 and the intention interpretation device 40 may be configured as an integrated device.

Although the embodiments of the present invention have been described above in detail with reference to the drawings, the specific configuration is not limited to these embodiments, and includes designs within the scope of the gist of the present invention.

100...Customer service business system, 10...Response control device, 20...Flow control device, 30...Speech recognition device, 40...Intention interpretation device, 50...Speech synthesis device, 60...Business execution device, 70...Business system, 80... Operator terminal, 200... User terminal, 21... Communication unit, 22... Call history storage unit, 23... Flow information storage unit, 25... Control unit, 251... Session control unit, 252... History control unit, 253... Voice recognition control unit , 254...Intention interpretation control unit, 255...Speech synthesis control unit, 256...Business control unit

Claims (5)

a control unit that determines a response sentence to be presented to the customer based on the content of the customer's utterance, determines the business content in response to the customer's request, and instructs a business execution device to execute the business content; Equipped with
The control unit selects a device that will be the main body for executing the process of analyzing the utterance content from among a plurality of devices capable of the same type of processing ;
The control unit selects an intention interpretation device that has relatively high accuracy in interpreting the intention of the complaint as the device that will be the execution subject when there is a high possibility that a voice indicating a complaint will be obtained from the customer; A control device that selects a speech synthesis device that satisfies predetermined conditions that are suitable for a complaint and obtains an execution result of speech synthesis processing . The content of the utterance is expressed in audio,
The control unit requests a speech recognition device that performs speech recognition processing to perform speech recognition processing of the speech indicating the content of the utterance, and determines the response sentence based on the recognition result obtained from the speech recognition device. , determine the business content,
2. The control unit selects, from among a plurality of speech recognition devices, a device that will be the main subject of execution of the speech recognition process as a device that satisfies the condition, if a predetermined condition is satisfied. control device. The control unit requests a speech synthesis device that performs speech synthesis processing to generate speech representing the response sentence, and responds to the customer using the speech obtained from the speech synthesis device.
3. The control unit selects, from a plurality of speech synthesis devices, a device that will be the main subject of execution of the speech synthesis process as a device that satisfies the condition, if a predetermined condition is satisfied. Control device. a first step of determining a response sentence to be presented to the customer based on the content of the customer's utterance;
a second step of determining the business content according to the customer's request;
a third step of instructing a business execution device to execute the business content;
a fourth step of selecting a device that will be the main body for executing any of the processes in the first step, second step, and third step from among a plurality of devices capable of performing the same type of processing;
In a situation where there is a high possibility that voice indicating a complaint will be obtained from the customer, an intention interpretation device with relatively high precision in interpreting the intention of the complaint is selected as the device that will be the main executioner, and the device is suitable for responding to the complaint. a fifth step of selecting a speech synthesizer that satisfies a predetermined condition and obtaining the execution result of the speech synthesis process;
A control method having A computer program for causing a computer to function as the control device according to claim 1.

Images