Automatic voice response method, automatic voice response program, and automatic voice response device

Abstract

This invention provides an automated voice response system, method, and program that enables high-quality conversations similar to those of a human operator by linking with an external purchasing information management server and automatically generating detailed responses based on the user's individual attributes and history. [Solution] In the reservation system, the automated voice response system receives inquiries through voice interaction with the user terminal, acquires user information from the user terminal, accesses an external purchase information management server based on the inquiry content and user information, acquires purchase-related data including user attribute information, purchase history, preference information, or interaction history, generates a response tailored to the individual user's situation from the acquired purchase-related data and inquiry content, and transmits it to the user terminal.

Description

Technical Field

[0001] The present invention relates to a voice automatic response method, a voice automatic response program, and a voice automatic response device.

Background Art

[0002] In recent years, dialogue systems using large language models (LLMs) have been put into practical use. In a large language model, response text is generated using a language model for a prompt input by text. However, by using the conversion between voice and text, it has become possible to respond to voice input from a user with voice by a large language model. Patent Document 1 discloses a dialogue system that uses an estimation model to estimate a user's intention from voice information or text information and conducts product selection, product purchase, or service reservation reception through voice dialogue.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the above conventional dialogue system, there is a problem that it is difficult to make a response based on the detailed situation and past history of an individual user. For example, even in a dialogue system using a large language model, what the model itself has learned is general knowledge, and it does not hold individual information such as "when and what was purchased", "what kind of preferences are there", and "what inquiries were made in the past" of a specific user.

[0005] Therefore, when users made inquiries based on their individual purchase history and attribute information, such as "I want to know the delivery status of the product I recently purchased," "I'm looking for something similar to a product I previously purchased," or "I want to confirm whether payment has been completed," the conventional automated response system could not generate an answer, and the call had to be transferred to a human operator. This failed to achieve the purpose of automated voice response, which is to reduce the workload of operators, and also detracted from the customer experience for users.

[0006] This invention has been made in view of the above circumstances, and its purpose is to provide an automated voice response method, program, and apparatus that can achieve high-quality dialogue similar to that of a human operator by cooperating with an external purchasing information management server and automatically generating detailed responses based on the user's individual attributes and history in an automated voice response system. [Means for solving the problem]

[0007] A voice automated response method performed by a computer connected via a network to a user terminal and a purchasing information management server, The steps include: receiving a voice call request from the user terminal via a predetermined telephone number and performing a voice conversation with the user of the user terminal; The steps include obtaining user information from the user terminal, The steps include receiving inquiries from the user through the aforementioned voice interaction, The steps include requesting purchase-related data from the purchase information management server based on the content of the inquiry and the user information, and receiving the purchase-related data from the purchase information management server, The steps include generating response text for the voice dialogue based on the received purchase-related data and the content of the inquiry, The steps include converting the response text into speech and sending it to the user terminal, A voice-activated automated response method that includes this. [Effects of the Invention]

[0008] According to the method, apparatus, and program of the present invention, it is possible to provide an automated voice response method, program, and apparatus that can achieve high-quality dialogue similar to that of a human by automatically generating detailed responses based on the user's individual attributes and history. [Brief explanation of the drawing]

[0009] [Figure 1] This is a block diagram showing the hardware configuration of the reservation system. [Figure 2] This is a block diagram showing the functional configuration of the automated voice response system 1. [Figure 3] This is a flowchart showing the processing flow for reservation acceptance. [Figure 4] This is a block diagram showing the hardware configuration of a reservation system in another embodiment. [Figure 5] This is a block diagram showing the functional configuration of the automated voice response system 1. [Figure 6] This is a flowchart showing the processing flow for reservation acceptance. [Figure 7] This is a block diagram showing the hardware configuration of the order receiving system. [Figure 8] This is a block diagram showing the functional configuration of the automated voice response system 1. [Figure 9] This is a flowchart showing the order processing flow. [Figure 10] This is a flowchart showing the order processing flow. [Modes for carrying out the invention]

[0010] (First Embodiment) Hereafter, identical or equivalent components, members, processes, and signals shown in each drawing will be denoted by the same reference numerals, and redundant explanations will be omitted as appropriate. In addition, some components that are not explanatoryly important will be omitted from each drawing. The following embodiments are not intended to unduly limit the contents of the disclosure as described in the claims, and not all components shown in the embodiments are necessarily essential components of the disclosure. Furthermore, each figure is a schematic diagram and is not necessarily a strict illustration.

[0011] (System Configuration) The reservation system described in this disclosure is a system that provides reservation acceptance via voice calls with users. The reservation system comprises a voice automated response device 1, a user terminal 2, and a reservation management server 3, all connected via a network NW. Figure 1 is a block diagram showing the hardware configuration of the reservation system. Figure 2 is a block diagram showing the functional configuration of the voice automated response device 1.

[0012] (Hardware configuration of the automated voice response device 1) The automated voice response device 1 comprises a control unit 101, a storage unit 102, and a communication unit 103. The control unit 101 is composed of a processor and reads application programs and data necessary for processing stored in the storage unit 102, executes processing, and sends and receives data to and from the user terminal 2 and reservation management server 3 connected to the network NW via the communication unit 103, thereby realizing the functions of the application program. In this way, the automated voice response device 1 is realized as an information processing device that performs information processing. The automated voice response device 1 is a computer installed on the network, but it may also be a server virtually configured as a so-called cloud, or it may be composed of terminals with communication functions such as personal computers and smartphones.

[0013] The memory unit 102 is a storage device for permanently or temporarily storing data and programs, and is composed of a volatile memory, a hard disk, an SSD, etc. The communication unit 103 communicates with other computers via a network NW according to wired or wireless communication standards. The network NW is composed of a mobile communication network, a wireless LAN, other short-range wireless communications, a telephone line, or a combination thereof, and the communication unit 103 is composed of an interface capable of inputting and outputting signals with the network for communication.

[0014] (Configuration of the control unit 101 of the voice automatic response device 1) The control unit 101 is composed of functional units of an information registration unit 1011, a reservation information acquisition unit 1012, a voice call unit 1013, a temporary reservation processing unit 1014, a final reservation processing unit 1015, and a notification unit 1016. The control unit 101 reads an application program stored in the memory unit 102 and executes the processing, whereby each functional unit is realized.

[0015] (Functional configuration of the memory unit 102 of the voice automatic response device 1) The memory unit 102 stores and holds an application program for providing a reservation reception function to the user terminal 2, user data, and service data. The application program is executed by a processor and operates as each functional unit of the control unit 101 to realize the reservation reception function. The user data is specifically a database of user attribute information, access information to the reservation management server 3, and the user's reservation status. The service data is data necessary for each functional unit of the control unit 101 to provide a reservation reception service, and specifically includes parameters used in the application program, screen data to be displayed on the user terminal 2, templates for transmission messages to the user, template data for prompts to be transmitted to a large language model (not shown), etc.

[0016] (Functional Configuration of Communication Unit 103 of Voice Automatic Response Device 1) The communication unit 103 provides an interface for communication in the operations of each functional unit of the control unit 101. When each functional unit of the control unit 101 communicates with the user terminal 2 and the reservation management server 3, the communication unit 103 mediates the communication with the network NW.

[0017] (Information Registration Unit 1011) The information registration unit 1011 stores the telephone numbers of service providers (hereinafter referred to as service providers) for which reservation reception is targeted in the storage unit 102. The storage unit 102 stores the telephone numbers of multiple service providers for each service provider. The information registration unit 1011 stores the access information of the reservation management server 3. Specifically, the access information is physical access information such as the IP address and URL of the reservation management server 3, the administrator user ID and password of the reservation management server 3, or the user ID and password information for each service provider. The administrator user of the reservation management server 3 can have the authority to access the information of each service provider. When having this authority, the access information does not need to include the user ID and password information for each service provider. When not having the authority, the user ID and password information for each service provider is stored in the storage unit 102. In this embodiment, the voice automatic response device 1 is described as performing subsequent processing after accessing the reservation management server 3 with the administrator user ID and password having the authority to access the information of each service provider. However, even when not having the authority, equivalent processing is performed by accessing with the user ID and password for each service provider.

[0018] Telephone numbers and access information are provided to the terminal of the reservation management system administrator or service provider via a screen from the information registration unit 1011. The administrator or service provider enters the telephone number and access information from that screen and sends it to the reservation management system. The information registration unit 1011 accepts the input and stores it in the storage unit 102. Although it is stated that the telephone numbers registered by the information registration unit 1011 are stored for each service provider, they may be a common number for all service providers and a telephone number that can be called by the automated voice response system 1.

[0019] (Reservation information acquisition unit 1012) The reservation information acquisition unit 1012 refers to the storage unit 102 and accesses the reservation management server 3 using the user ID and password as an administrator user to acquire reservation status information for each service provider. The reservation status is acquired for a predetermined period for all service providers whose services are subject to reservation acceptance by the reservation acceptance system. In this embodiment, the target period is 3 months, but it may be any period depending on the service content and industry of the service provider, or the reservation acceptance system may set a limit on the period during which reservations are accepted and use that period. The reservation information acquisition unit 1012 stores the acquired reservation status information in the reservation acceptance database of the storage unit 102.

[0020] The information stored in the reservation database includes records containing the service provider's identification information, the service's identification information, the date and time, the status, and the reservation holder's information. The status consists of three states: unbooked, tentatively booked, and confirmed booked. The reservation information acquisition unit 1012 acquires information on available date and time slots for the service from the reservation management server 3, and creates a new record in the reservation database with the service provider's identification information and the service's identification information, setting the status to unbooked and the reservation holder's information to undecided.

[0021] The reservation information acquisition unit 1012 acquires reservation status from the reservation management server 3 at predetermined time intervals. This is intended to synchronize the reservation acceptance database in the storage unit 102 of the automated voice response device 1 with the information from the reservation management server 3 at regular time intervals, and is always performed asynchronously and independently of other processing by the automated voice response device 1. The predetermined time interval is preferably within 5 seconds, but while a shorter interval can prevent duplicate reservations, a shorter interval also has disadvantages such as an increased processing load on the automated voice response device 1 and the reservation management server 3, so it is determined as appropriate depending on the content of the service, the timing, and the environment.

[0022] (Voice call unit 1013) The voice call unit 1013 receives a voice call request from the user terminal 2 via the communication unit 103 and performs a voice call with the user. The user dials the telephone number of the service provider they wish to book from the user terminal 2, and the communication unit 103 of the automated voice response device 1 rings the call using that telephone number. The communication unit 103 is configured to ring using the telephone numbers of each service provider stored in the storage unit 102. After the call is made, the automated voice response device 1 performs an automated response call with the user terminal via the voice call unit 1013.

[0023] The voice call unit 1013 generates call text by converting the call audio from the user terminal 2 into text using the speech recognition function of a large-scale language model. The large-scale language model generates response text to the user from the call text and prompts in the storage unit 102, converts it into speech using the text-to-speech function to generate response audio, and transmits the response audio via the communication unit 103. The voice call unit 1013 performs automatic voice calls with the user terminal 2 by repeating this series of processes.

[0024] The large-scale language model is accessed via an API from an external large-scale language model provider server connected by a network. Alternatively, the large-scale language model may be stored as an executable program in the storage unit 102 and executed within the voice automatic response device 1.

[0025] The prompts given to the large-scale language model include instructions such as: "1. Ask the user in a dialogue format about the purpose of their call," "2. If the purpose is a reservation, ask in a dialogue format about the user's information, such as their name and contact information, and reservation request information, such as the content and time of the desired service," and "3. Refer to the provided information for information about the service content and service provider." The provided information consists of information related to the service provider corresponding to the incoming telephone number, stored in the memory unit 102.

[0026] (Provisional reservation processing 1014) The provisional reservation processing unit 1014 checks the availability of the service by referring to the reservation acceptance database in the storage unit 102 for the reservation request information received by the voice call unit 1013. If the status of the record corresponding to the reservation request information is "not reserved", the provisional reservation processing unit 1014 updates the status in the corresponding reservation acceptance database to "provisional reservation by the user" and updates the reservation holder information to user information. If the status of the record corresponding to the reservation request information is "unavailable (provisional reservation, confirmed reservation)", the provisional reservation processing unit 1014 notifies the voice call unit 1013 of this fact, and the voice call unit 1013 continues the automated voice call with the user and obtains the reservation request information in a dialogue format.

[0027] The provisional reservation processing unit 1014 notifies the user via the voice call unit 1013 that the reservation corresponding to the user's reservation request information has been made a provisional reservation. Specifically, the voice call unit 1013 generates a response voice in response to the notification and informs the user terminal 2 via voice that the provisional reservation has been accepted. Furthermore, the voice call unit 1013 conveys a message indicating that the reservation is complete and ends the voice call with the user terminal 2.

[0028] (This reservation processing step 1015) The reservation processing unit 1015 refers to the storage unit 102 and accesses the reservation management server 3 using the user ID and password as an administrator user. For records in the reservation acceptance database whose status is "tentative reservation," the reservation processing unit 1015 requests a reservation from the reservation management server 3. If the reservation is successful on the reservation management server 3, the reservation processing unit 1015 updates the status of the corresponding record in the reservation acceptance database to "confirmed reservation" and notifies the notification unit 1016 accordingly.

[0029] The reservation processing unit 1015 makes reservation requests from the reservation management server 3 at predetermined time intervals. This is intended to synchronize the reservation status on the reservation management server 3 with records in the reservation acceptance database within the storage unit 102 of the automated voice response device 1 whose status is "provisional reservation," and is to ensure that it is executed asynchronously and independently of other processes of the automated voice response device 1. If the reservation processing unit 1015 fails on the reservation management server 3 due to network communication conditions or other reasons, the status in the reservation acceptance database is not updated. As a result, the reservation request processing on the reservation management server 3 is re-executed at predetermined time intervals even for the failed records.

[0030] (Notification Department 1016) The notification unit 1016 notifies the user terminal 2 that the status in the reservation acceptance database has become a confirmed reservation. Since the notification occurs after the voice call with the user has ended, it is sent as a short message to the phone number that called the user terminal 2. If the user's email address or SNS recipient information is registered in the storage unit 102, notification may also be sent via those means.

[0031] (Configuration of User Terminal 2) User terminal 2 comprises a control unit, a storage unit, and a communication unit. The processor of user terminal 2 functions as the control unit by calling and executing application programs from the storage unit. The storage unit stores data necessary for the execution of application programs by the processor of user terminal 2. The communication unit provides the interface function for when user terminal 2 communicates via the network NW.

[0032] The hardware configuration of user terminal 2 is preferably a mobile communication terminal such as a smartphone with calling capabilities, but any terminal with calling capabilities, such as a tablet, laptop computer, or landline telephone, may also be used. The hardware components that make up each function of user terminal 2 are as follows: the control unit consists of a processor, and the storage unit consists of volatile memory, a hard disk, or an SSD. The communication unit consists of a communication interface to a mobile communication network, wireless LAN, mobile phone line, landline telephone line, or a combination thereof. The display unit is a liquid crystal display screen, preferably configured with an overlaid touch panel that also serves as an input means.

[0033] (Configuration of reservation management server 3) The reservation management server 3 is a server that manages the reservation status of each service provider in association with the user who made the reservation, using a database. The reservation management server 3 is connected to the automated voice response system 1 via a network (NW), and communicates with the automated voice response system 1 using a web interface, API, or any combination of other proprietary protocols of this reservation management system, to accept reservation requests. Although not shown in Figure 1, the reservation management server 3 is connected to multiple automated voice response systems 1, and can accept and centrally manage reservation requests from multiple automated voice response systems 1. This distributes the concentration of incoming calls on the automated voice response system 1, and even if an automated voice response system 1 is down, calls can still be made by other automated voice response systems 1, preventing situations where users cannot make calls.

[0034] (Operation of the automated voice response system 1) The following describes each process of the automated voice response system 1. Figure 3 is a flowchart of the process operation.

[0035] Referring to Figure 3, the flow of the reservation acceptance process performed by the automated voice response system 1 according to this embodiment will be explained. Although the series of processes in this flowchart are performed by the control unit 101 of the automated voice response system 1, for the sake of clarity, the functional unit that is the main component of the process in each step will be explicitly indicated and explained.

[0036] (Step S01) The information registration unit 1011 stores the telephone numbers of service providers whose reservations are accepted and the access information for the reservation management server 3. Specifically, the access information includes physical access information such as the IP address and URL of the reservation management server 3, as well as the administrator user ID and password for the reservation management server 3, or the user ID and password for each service provider.

[0037] Telephone number and access information are provided to the terminal of the reservation management system administrator or service provider via a screen from the information registration unit 1011. The administrator or service provider enters the telephone number and access information from that screen and sends it to the reservation management system. The information registration unit 1011 accepts the input and stores it in the storage unit 102.

[0038] (Step S02) Next, the reservation information acquisition unit 1012 acquires reservation information. Specifically, the reservation information acquisition unit 1012 accesses the reservation management server 3 at a first predetermined time interval, acquires the latest reservation status, and updates the reservation acceptance database in the storage unit 102. In Figure 3, the loop (repetition) arrow next to S02 indicates that this process is executed repeatedly in the background at the first predetermined time interval, regardless of whether or not there is an incoming call from a user. In other words, the process in step S02 is executed asynchronously and independently of the other steps. As a result, the reservation acceptance database in the voice automated response device 1 is always synchronized with the reservation management server 3.

[0039] (Step S03) When a call is made from user terminal 2 to the service provider's phone number, the voice call unit 1013 receives the call and a voice call is initiated. The voice call unit 1013 uses a large-scale language model such as a generative AI to engage in natural language dialogue with the user.

[0040] (Step S04) The voice call unit 1013 engages in natural language dialogue with the user using a large-scale language model and receives reservation request information such as date and time, number of people, menu, and service details.

[0041] (Step S05) Upon receiving reservation request information, the provisional reservation processing unit 1014 refers to the reservation acceptance database in the storage unit 102 to check availability. If the status of the record corresponding to the reservation request information is "not reserved," the provisional reservation processing unit 1014 updates the status in the corresponding reservation acceptance database to "provisional reservation by the user" and the reservation holder information to the user information to prevent duplicate reservations from other users. If the status of the record corresponding to the reservation request information is "unavailable (provisional reservation, confirmed reservation)," the provisional reservation processing unit 1014 notifies the voice call unit 1013 of this fact, and the voice call unit 1013 continues the automated voice call with the user and obtains the reservation request information in a dialogue format. At this time, instead of querying the external reservation management server 3, the local database updated in step S02 is referred to, so availability can be determined immediately without being affected by network communication delays, timeouts, or the response of the reservation management server 3.

[0042] (Step S06) The voice call unit 1013 sends a provisional reservation notification to the user. For example, it outputs a voice message such as, "We have received your provisional reservation for your requested date and time. We will contact you via SMS once it is confirmed," and then ends the call. This completes the voice call processing with the user.

[0043] (Step S07) After the call ends, or even during the call, the reservation processing unit 1015 performs the reservation processing in parallel. The reservation processing unit 1015 monitors the reservation acceptance database at a second predetermined time interval, and when it detects a record with the status "tentative reservation", it performs reservation processing on the reservation management server 3 based on the information of the corresponding record.

[0044] (Step S08) Once the reservation process to the reservation management server 3 is successful and the reservation is confirmed, the reservation processing unit 1015 updates the status of the reservation acceptance database.

[0045] (Step S09) The notification unit 1016 sends this reservation notification to the user terminal 2. This notification is sent via a push-type communication method such as SMS (Short Message Service) or email. As a result, the user will receive a reservation completion notification a short time after hanging up the phone. With these steps, the series of reservation acceptance processes is completed.

[0046] (Asynchronous execution of this reservation process) In Figure 3, the loop arrow from step S07 to step S09 indicates that it is executed asynchronously and repeatedly at predetermined time intervals, independently of the processing of other steps. In step S07, the reservation acceptance database is monitored at predetermined time intervals, and based on the results, steps S08 and S09 are executed, resulting in steps S07 to S09 being executed repeatedly at predetermined time intervals. In the flowchart of Figure 3 and in the description of this embodiment, steps S07 to S09 are conveniently described separately, but it is also possible to consider the series of processes consisting of steps S07 to S09 as a single step, and to understand that this step is executed asynchronously and repeatedly at predetermined time intervals, independently of the processing of other steps.

[0047] The predetermined time interval in step S07 (the second predetermined time interval) aims to synchronize the reservation acceptance database and the reservation management server 3. If there is one automated voice response device 1 connected to one reservation management server 3, the second predetermined time interval may be longer than the predetermined time interval in step S02 (for example, several tens of minutes or more). However, if there are multiple automated voice response devices 1 connected to one reservation management server 3, it is necessary to maintain consistency between the databases of the automated voice response devices 1, so the second predetermined time interval must be at least the same as the first predetermined time interval or shorter. In other words, the second predetermined time interval is set appropriately depending on the first predetermined time interval and the configuration of the automated voice response system.

[0048] Please note that the flowchart in Figure 3 is just one example, and is presented in this chronological order for clarity. However, the representation in this diagram and the explanation of the processing flow described above are not necessarily the only valid approaches. Steps S02 and S07-S09 are independently and asynchronously repeated at predetermined time intervals, operating as so-called background processes. This is also true for subsequent embodiments.

[0049] As explained above, according to this embodiment, during voice interaction (S03-S06), there is no access to the external reservation management server 3, and processing is performed by referring to the local database. Therefore, it is not affected by response delays from the external server, enabling smooth voice interaction. Furthermore, by performing the reservation processing to the external server (S07-S09) asynchronously, even if the external server is undergoing maintenance or is under heavy load, the acceptance of reservations from users can continue without interruption, thus improving the robustness of the system.

[0050] (Second embodiment) In the first embodiment, the automated voice response system 1 received reservation information via voice calls. In this embodiment, it further has the function of responding to user inquiries via voice calls. This makes it possible to provide service explanations, recommendations based on purchase history, inquiries based on response history, payment processing, and cancellation processing when handling calls with users. Note that this embodiment has many components in common with the previous embodiment, so the differences will be explained in detail.

[0051] (System Configuration) In this embodiment, the reservation system includes a purchasing information management server 4 in addition to the configuration shown in Figure 1, as shown in Figure 4. The purchasing management server 4 is configured to communicate with the voice automated response device 1 via a network NW. The purchasing management server 4 is preferably configured separately from the voice automated response device 1, either as a standalone server or a server implemented in the cloud, from the standpoint of load balancing and fault avoidance, but it may also be configured together with the voice automated response device 1 or the reservation management server 3.

[0052] (Configuration of Purchasing Management Server 4) The Purchase Information Management Server 4 is a server device that integrates sales management functions for managing the sales history of products and services with customer relationship management (CRM) functions for managing user attributes and preferences. The Purchase Information Management Server 4 has a "customer database" and a "product database" in its storage unit. The Purchase Information Management Server 4 has functions to manage payments, inquiries, and purchase history for reservations that have been confirmed by the Reservation Management Server 3, in cooperation with the Reservation Management Server 3. The Purchase Information Management Server 4 also has functions to manage information on products and services sold.

[0053] The "Customer Database" uses user information such as the phone number of User Terminal 2 as a key to store basic attributes of the user, such as their name, address, and email address, as well as past purchase history (purchase date and time, purchased items, amount), inquiry history, and user preference information analyzed by CRM functions. The "Product Database" manages detailed information about the products and services handled (name, price, specifications, stock status) and information on recommended products related to each product (products for upselling and crossselling).

[0054] The automated voice response system 1 can use the API function of the purchase information management server 4 to obtain purchase-related data consisting of user attribute information, purchase history, inquiry history, user preference information, detailed information on the services it handles, information such as recommended products related to each product, and payment information for products and services, from user information such as telephone numbers and user inquiries obtained through voice calls.

[0055] (Configuration of the control unit 101 of the automated voice response device 1) In this embodiment, the functional configuration of the control unit 101 of the voice automatic response device 1 is further enhanced by the configuration of the first embodiment, including a purchase-related data acquisition unit 1017 and an inquiry handling unit 1018, as shown in Figure 5.

[0056] (Purchase-related data acquisition unit 1017) The purchase-related data acquisition unit 1017 requests purchase-related data from the purchase information management server 4 via the communication unit 103, using the API function, based on the user's telephone number obtained by the voice call unit 1013 and the user's inquiry content obtained through the voice call. The key information when requesting purchase-related data via API access includes the user's attribute information (name, address, etc.) included in the inquiry content, and information about the product and service being inquired about. The user's purchase-related data consists of data such as the user's attribute information, purchase history, inquiry history, user preference information, detailed information about the products and services handled, information such as recommended products related to each product, and payment information for products and services. The purchase-related data acquisition unit 1017 stores the acquired purchase-related data in the storage unit 102.

[0057] (Inquiry Support Department 1018) The inquiry handling unit 1018 retrieves purchase-related data from the storage unit 102 related to the user whose inquiry is being handled. The inquiry handling unit 1018 retrieves the call text from the voice call unit 1013 and sends the retrieved call text, purchase-related data, and a prompt instructing the inquiry handling to the large-scale language model via the communication unit 103. The large-scale language model is an external large-scale language model provider server connected by a network NW, and the data is sent via an API.

[0058] The prompts that the inquiry response unit 1018 sends to the large-scale language model include instructions such as: "1. Respond to the user's inquiry in a conversational format," "2. Generate the response by referring to the interaction history in the purchase-related data for tone, conversation content, etc.," "3. If the inquiry is about which product or service to choose, generate a response that includes recommendations by referring to the product information in the purchase-related data," and "4. If the inquiry concerns a procedure, refer to the purchase history and payment information in the purchase-related data to generate a response about the procedure, and if the user needs to perform the procedure, include in the response a statement that the information for that will be sent separately as a text message. Also output the information necessary to send the text message."

[0059] The inquiry handling unit 1018 receives the response text of the prompt sent from the large-scale language model, sends the response text to the voice call unit 1013, and the voice call unit 1013 converts the response text into speech and sends it to the user terminal 2, providing a voice call response. The inquiry handling unit 1018 continues the voice call until the inquiry handling with the user is completed, at which point it terminates the voice call.

[0060] (Operation of the automated voice response system 1) The following describes each process of the automated voice response device 1 in this embodiment. Figure 6 is a flowchart of the process operation. The difference from the process flow shown in Figure 3 in the first embodiment is steps S10 and S11.

[0061] Referring to Figure 6, the flow of the reservation acceptance process performed by the automated voice response system 1 according to this embodiment will be described. In this embodiment, only steps S10 and S11 will be described. The processing of the other steps is the same as in the first embodiment and will therefore be omitted.

[0062] (Step S10) In step S03, the voice call unit 1013 engages in natural language dialogue with the user using a large-scale language model such as a generative AI. As previously mentioned, the prompt given to the large-scale language model includes the instruction "1. Ask the user about the purpose of the call in a dialogue format." In the first embodiment, the process was described assuming that the purpose was a reservation. In this embodiment, a determination is made as to whether the purpose is a reservation, and the processing steps are branched accordingly. If the purpose of the call is a reservation, the process proceeds to step S04; otherwise, the process proceeds to step S11.

[0063] (Step S11) The purchase-related data acquisition unit 1017 requests purchase-related data from the purchase information management server 4 via the communication unit 103, using the API function, based on the user's telephone number obtained by the voice call unit 1013 and the user's inquiry content obtained through the voice call. The purchase-related data consists of data such as user attribute information, purchase history, inquiry history, user preference information, detailed information on products and services handled, information such as recommended products related to each product, and payment information for products and services. The purchase-related data acquisition unit 1017 stores the acquired purchase-related data in the storage unit 102.

[0064] (Step S12) The inquiry handling unit 1018 retrieves purchase-related data from the storage unit 102 related to the user whose inquiry is being handled. The inquiry handling unit 1018 retrieves the call text from the voice call unit 1013 and sends the retrieved call text, purchase-related data, and a prompt instructing the inquiry handling to the large-scale language model via the communication unit 103. The large-scale language model is an external large-scale language model provider server connected by a network NW, and the data is sent via an API. The prompt to be sent is as described above.

[0065] The inquiry handling unit 1018 receives the response text of the prompt sent from the large-scale language model, sends the response text to the voice call unit 1013, and the voice call unit 1013 converts the response text into speech and sends it to the user terminal 2, performing a voice call response. The inquiry handling unit 1018 continues the voice call until the voice call response to the inquiry with the user is completed, at which point the voice call is terminated. Although not shown in the flowchart in Figure 6, if the user wishes to make a reservation for the service during the inquiry response, the process proceeds to step S04 to obtain the reservation request information and perform the reservation acceptance process. The flow of the reservation acceptance process is the same as in the first embodiment.

[0066] As described above, according to this embodiment, the automated voice response device 1 works in conjunction with the purchase information management server 4 to acquire purchase-related data, including user attributes, purchase history, and preference information, using the user's telephone number as a key. This individual purchase-related data and the inquiry content are then input into a large-scale language model to generate a response. This enables not only a simple, standardized automated response, but also a detailed response similar to that of a human, taking into account the user's past history and preferences (product recommendations, handling inquiries based on past transactions, procedural guidance, etc.). As a result, users can be provided with a one-stop, high-quality voice interaction experience for a variety of inquiries, not just reservation requests.

[0067] (Third embodiment) The first and second embodiments described reservation management systems for service reservations. This embodiment describes an order acceptance system with a configuration similar to the first and second embodiments, which takes orders for products via telephone calls to users and provides real-time and robustness for product inventory confirmation and order management.

[0068] (System Configuration) The order taking system described in this disclosure is a system that provides the acceptance of product orders via voice calls with users. The order taking system comprises a voice automated response device 1, a user terminal 2, and an order management server 5, all connected via a network NW. Figure 7 is a block diagram showing the hardware configuration of the order taking system. Figure 8 is a block diagram showing the functional configuration of the voice automated response device 1. The following description will be given with reference to Figures 7 to 9.

[0069] (Hardware configuration of the automated voice response device 1) The automated voice response device 1 comprises a control unit 101, a storage unit 102, and a communication unit 103. The control unit 101 is composed of a processor and reads application programs and data necessary for processing stored in the storage unit 102, executes processing, and sends and receives data to and from the user terminal 2 and order management server 5 connected to the network NW via the communication unit 103, thereby realizing the functions of the application program. In this way, the automated voice response device 1 is realized as an information processing device that performs information processing. The automated voice response device 1 is a computer installed on the network, but it may also be a server virtually configured as a so-called cloud, or it may be composed of terminals with communication functions such as personal computers and smartphones.

[0070] The memory unit 102 is a storage device for permanently or temporarily storing data and programs, and is composed of volatile memory, hard disks, SSDs, etc. The communication unit 103 communicates with other computers via a network NW using wired or wireless communication standards. The network NW is composed of a mobile communication network, wireless LAN, other short-range wireless communication, telephone lines, or a combination thereof, and the communication unit 103 is composed of an interface that can communicate with the network by inputting and outputting signals.

[0071] (Configuration of the control unit 101 of the automated voice response device 1) The control unit 101 is composed of functional units: 1011 Information Registration Unit, 1019 Inventory Information Acquisition Unit, 1013 Voice Communication Unit, 1020 Provisional Order Processing Unit, 1021 Final Order Processing Unit, and 1016 Notification Unit. Each functional unit is realized by the control unit 101 reading application programs stored in the storage unit 102 and executing processing.

[0072] (Functional configuration of the memory unit 102 of the automated voice response device 1) The memory unit 102 stores and maintains application programs, user data, and product data for providing order acceptance functionality to the user terminal 2. The application programs are executed by the processor and operate as various functional units of the control unit 101 to realize the order acceptance functionality. The order acceptance database built in the memory unit 102 has an "inventory management table" that manages the inventory quantity for each product and an "order management table" that manages the details of individual orders from users. User data specifically includes user attribute information, access information to the order management server 5, and a database of the user's order status. Product data is data necessary for each functional unit of the control unit 101 to provide order acceptance services, and specifically includes parameters used in the application programs, screen data to be displayed on the user terminal 2, templates for messages sent to the user, and template data for prompts sent to a large-scale language model (not shown).

[0073] (Functional configuration of the communication unit 103 of the automated voice response device 1) The communication unit 103 provides an interface for communication during the operation of each functional unit of the control unit 101. When each functional unit of the control unit 101 communicates with the user terminal 2 and the order management server 5, the communication unit 103 acts as an intermediary for communication with the network NW.

[0074] (Information Registration Section 1011) The information registration unit 1011 stores the telephone numbers of the sellers of the products subject to order acceptance (hereinafter referred to as "product sellers") in the storage unit 102. The storage unit 102 stores the telephone numbers of multiple product sellers, one for each product seller. The information registration unit 1011 also stores access information for the order management server 5. Specifically, the access information includes physical access information such as the IP address and URL of the order management server 5, and the administrator user ID and password for the order management server 5, or the user ID and password for each product seller. The administrator user of the order management server 5 can have the authority to access the information of each product seller, and if this authority is present, the access information does not need to include the user ID and password for each product seller. If the authority is not present, the user ID and password for each product seller is stored in the storage unit 102. In this embodiment, the automated voice response device 1 is described as accessing the order management server 5 using an administrator user ID and password that grants access to each product seller's information, and then proceeding with the subsequent processing. However, even if the user does not have the necessary permissions, the same processing can be performed by accessing the server using each product seller's user ID and password.

[0075] Telephone numbers and access information are provided to the terminals of the order management system administrator or product seller via a screen from the information registration unit 1011. The order management system administrator or product seller enters the telephone number and access information on that screen and sends it to the order management system. The information registration unit 1011 accepts the input and stores it in the storage unit 102. Although it is stated that the telephone numbers registered by the information registration unit 1011 are stored for each product seller, they may be a common number for all product sellers and a telephone number that can be called by the automated voice response system 1.

[0076] (Inventory Information Acquisition Unit 1019) The inventory information acquisition unit 1019 refers to the storage unit 102 and accesses the order management server 5 using the user ID and password as an administrator user to acquire inventory information for each product seller. It acquires inventory information for a predetermined period or for predetermined items from all sellers of products that are subject to order acceptance by the order acceptance system. In this embodiment, the target period and items are all products handled, but it may be any product category depending on the industry and sales strategy of the product seller, or the order acceptance system may set restrictions on the products for which orders are accepted and limit the scope of the inventory information acquisition unit. The inventory information acquisition unit 1019 stores the acquired inventory information in the order acceptance database of the storage unit 102.

[0077] As mentioned above, the order acceptance database includes an "inventory management table" that manages the stock quantity for each product and an "order management table" that manages the details of individual orders from users. When the inventory information acquisition unit 1019 acquires inventory information from each product seller, it creates or updates records in the inventory management table for each piece of inventory information, including the product provider's identification information, product identification information, and stock quantity.

[0078] The inventory information acquisition unit 1019 acquires inventory information from the order management server 5 at predetermined time intervals. This is intended to synchronize the order acceptance database in the storage unit 102 of the automated voice response device 1 with the information from the order management server 5 at regular time intervals, and is performed asynchronously and continuously, independently of other operations by the automated voice response device 1. The predetermined time interval is preferably within 5 seconds, but while shorter intervals can prevent duplicate orders, shorter intervals also have disadvantages such as increased processing load on the automated voice response device 1 and the order management server 5, so it is determined as appropriate depending on the service content, timing, and environment.

[0079] (Voice Call Unit 1013) The voice call unit 1013 receives a voice call request from the user terminal 2 via the communication unit 103 and performs a voice call with the user. The user dials the telephone number of the seller of the product they wish to purchase from the user terminal 2, and the communication unit 103 of the automated voice response device 1 receives the call via that telephone number. The communication unit 103 is configured to receive calls via the telephone numbers of each product seller stored in the storage unit 102. After receiving the call, the automated voice response device 1 performs an automated response call with the user terminal via the voice call unit 1013.

[0080] The voice call unit 1013 generates call text by converting the call audio from the user terminal 2 into text using the speech recognition function of a large-scale language model. The large-scale language model generates response text to the user from the call text and prompts in the storage unit 102, converts it into speech using the text-to-speech function to generate response audio, and transmits the response audio via the communication unit 103. The voice call unit 1013 performs automatic voice calls with the user terminal 2 by repeating this series of processes.

[0081] The large-scale language model is accessed via an API from an external large-scale language model provider server connected by a network. Alternatively, the large-scale language model may be stored as an executable program in the storage unit 102 and executed within the voice automatic response device 1.

[0082] The prompts given to the large-scale language model include instructions such as: "1. Ask the user in a dialogue format about the purpose of their call," "2. If the purpose is to purchase a product, ask in a dialogue format about the desired product details, quantity, and other order information," and "3. Refer to the provided information for product details and seller information." The provided information is information from the storage unit 102 related to the seller corresponding to the incoming telephone number.

[0083] (Provisional Order Processing Unit 1020) The provisional order processing unit 1020 checks the available stock quantity of the product from the product seller by referring to the inventory management table in the order acceptance database in the storage unit 102, based on the order request information received by the voice call unit 1013. If the stock information indicates that the product is available for purchase (the stock quantity is greater than or equal to the required quantity), the provisional order processing unit 1020 updates the stock quantity of the product in the inventory management table by subtracting the number of orders, and generates a record in the corresponding order management table that includes the product seller's identification information, product identification information, status, number of orders, and customer information. At this time, the status is provisional order, and the customer information is based on user information. If there is no stock corresponding to the order request information and the order cannot be placed, the provisional order processing unit 1020 notifies the voice call unit 1013 of this fact, and the voice call unit 1013 continues the automated voice call with the user and obtains other order request information in a dialogue format.

[0084] The provisional order processing unit 1020 notifies the voice call unit 1013 that an order matching the user's order request information has been provisionally accepted. The voice call unit 1013 generates a response voice message in response to the notification and informs the user terminal 2 that the provisional order has been accepted. When notifying the user of the provisional order, the voice call unit 1013 conveys a message indicating that the order acceptance has ended and terminates the voice call with the user terminal 2.

[0085] (Order Processing Unit 1021) The order processing unit 1021 refers to the storage unit 102 and accesses the order management server 5 using the user ID and password as an administrator user. For records in the order management table of the order acceptance database whose status is provisional, the order processing unit 1021 requests the order management server 5 to accept the order. If the order is successful on the order management server 5, the order processing unit 1021 updates the status of the corresponding record in the order management table to confirmed order and notifies the notification unit 1016 of this fact.

[0086] Order requests from the order management server 5 by the order processing unit 1021 are made at predetermined time intervals. This is intended to synchronize the order status on the order management server 3 with records in the order management table of the order acceptance database in the storage unit 102 of the voice automated response device 1 whose status is "provisional order," and is to be executed asynchronously and independently of other processes of the voice automated response device 1. If the order processing by the order processing unit 1021 on the order management server 5 fails due to network communication conditions, etc., the status in the order management table will not be updated. As a result, the order request processing on the order management server 5 will be re-executed at predetermined time intervals even for the failed records.

[0087] (Notification Unit 1016) The notification unit 1016 notifies the user terminal 2 that the status of the order acceptance database has become confirmed as an order. Since the notification occurs after the voice call with the user has ended, it is sent as a short message to the phone number that called the user terminal 2. If the user's email address or SNS recipient information is registered in the storage unit 102, notification may also be made by those means.

[0088] (Configuration of User Terminal 2) User terminal 2 comprises a control unit, a storage unit, and a communication unit. The processor of user terminal 2 functions as the control unit by calling and executing application programs from the storage unit. The storage unit stores data necessary for the execution of application programs by the processor of user terminal 2. The communication unit provides the interface function for when user terminal 2 communicates via the network NW.

[0089] The hardware configuration of user terminal 2 is preferably a mobile communication terminal such as a smartphone with calling capabilities, but any terminal with calling capabilities, such as a tablet, laptop computer, or landline telephone, may also be used. The hardware components that make up each function of user terminal 2 are as follows: the control unit consists of a processor, and the storage unit consists of volatile memory, a hard disk, or an SSD. The communication unit consists of a communication interface to a mobile communication network, wireless LAN, mobile phone line, landline telephone line, or a combination thereof. The display unit is a liquid crystal display screen, preferably configured with an overlaid touch panel that also serves as an input means.

[0090] (Configuration of Order Management Server 5) The order management server 5 is a server that manages the inventory status of each product seller in association with the user who placed the order, using a database. The order management server 5 is connected to the automated voice response system 1 via a network (NW) and communicates with the automated voice response system 1 using a web interface, API, or any combination of other proprietary protocols of this product sales management system to accept orders. Although not shown in Figure 7, the order management server 5 is connected to multiple automated voice response systems 1, and can accept and centrally manage order requests from multiple automated voice response systems 1. This distributes the concentration of incoming calls to the automated voice response system 1, and even if one automated voice response system 1 is down, other automated voice response systems 1 can still accept calls, preventing situations where users cannot make calls.

[0091] (Operation of the automated voice response system 1) The following describes each process of the automated voice response system 1. Figure 9 is a flowchart showing the operation of the processes.

[0092] Referring to Figure 9, the flow of the order acceptance process performed by the automated voice response system 1 according to this embodiment will be explained. Although the series of processes in this flowchart are performed by the control unit 101 of the automated voice response system 1, for the sake of clarity, the functional unit that is the main component of the process at each step will be explicitly indicated and explained.

[0093] (Step S21) The information registration unit 1011 stores the telephone numbers of product sellers whose orders are being accepted and the access information for the order management server 5. Specifically, the access information includes physical access information such as the IP address and URL of the order management server 5, as well as the administrator user ID and password for the order management server 5, or the user ID and password for each product seller.

[0094] Telephone number and access information are provided to the terminal of the order management system administrator or product seller via a screen from the information registration unit 1011. The order management system administrator or product seller enters the telephone number and access information from that screen and sends it to the order management system. The information registration unit 1011 accepts the input and stores it in the storage unit 102.

[0095] (Step S22) Next, the inventory information acquisition unit 1019 acquires inventory information. Specifically, the inventory information acquisition unit 1019 accesses the order management server 5 at a first predetermined time interval, specifically, every few seconds, to acquire the latest inventory status and update the inventory management table in the order acceptance database of the storage unit 102. The loop arrow at S22 in Figure 9 indicates that this process is executed repeatedly in the background at the first predetermined time interval, regardless of whether or not there is an incoming call from a user. In other words, the process in step S22 is executed asynchronously and independently of the other steps. As a result, the order acceptance database in the voice automated response device 1 always holds up-to-date inventory information.

[0096] (Step S23) When a call is made from user terminal 2 to the product seller's phone number, the voice call unit 1013 receives the call and a voice call is initiated. The voice call unit 1013 uses a large-scale language model such as a generative AI to engage in natural language dialogue with the user.

[0097] (Step S24) The voice call unit 1013 engages in natural language dialogue with the user using a large-scale language model and receives order request information such as product name, quantity, and desired delivery date and time.

[0098] (Step S25) Upon receiving order request information, the provisional order processing unit 1020 refers to the inventory management table in the order acceptance database within the storage unit 102 to check the inventory status. If inventory is available, the provisional order processing unit 1020 updates each table in the order acceptance database. Specifically, it updates the inventory quantity of the product in the inventory management table by subtracting the number of orders, and creates a new record in the order management table to store product seller identification information, product identification information, and customer information, and sets the status of the record to "provisional order". This simultaneously secures inventory and records order information, preventing duplicate orders (overselling) from other users. At this time, instead of querying the external order management server 5, it refers to the local database updated in step S22, so inventory determination can be made immediately without being affected by network communication delays, timeouts, or the response of the order management server 5.

[0099] (Step S26) The voice call unit 1013 sends a provisional order notification to the user. For example, it outputs a voice message such as, "We have secured the product you requested. We will contact you by text message as soon as your order is confirmed," and then ends the call. This completes the synchronous dialogue process with the user.

[0100] (Step S27) After the call ends, or even during the call, the order processing unit 1021 performs order processing in parallel. The order processing unit 1021 monitors the order management table in the order acceptance database at a second predetermined time interval, and when it detects a record with the status "provisional order", it processes an order request to the order management server 5.

[0101] (Step S28) Once the write to the order management server 5 is successful and the order is confirmed, the order processing unit 1021 updates the status of the order acceptance database. It updates the status of the corresponding record in the order management table from "provisional order" to "confirmed order".

[0102] (Step S29) The notification unit 1016 sends an order confirmation to the user terminal 2. This notification is sent via a push-type communication method such as SMS (Short Message Service) or email. As a result, the user will receive an order completion notification a short time after hanging up the phone. With these steps, the order acceptance process is completed.

[0103] (Asynchronous execution of this order processing) In Figure 9, the loop arrow from step S27 to step S29 indicates that it is executed asynchronously and repeatedly at predetermined time intervals, independently of the processing of other steps. In step S27, the order acceptance database is monitored at predetermined time intervals, and based on the results, steps S28 and S29 are executed, resulting in steps S27 to S29 being executed repeatedly at predetermined time intervals. In the flowchart of Figure 9 and the description of this embodiment, steps S27 to S29 are conveniently described separately, but the series of processes consisting of steps S27 to S29 can be considered as a single step, and this step can be understood as being executed asynchronously and repeatedly at predetermined time intervals, independently of the processing of other steps. This means that even if access to the order management server 5 fails in step S27 due to congestion or other reasons, it will be executed after a predetermined time as a retry process for the relevant record or as a periodic batch process.

[0104] The predetermined time interval in step S27 (the second predetermined time interval) aims to synchronize the order acceptance database and the order management server 5. If there is only one automated voice response device 1 connected to one order management server 5, the second predetermined time interval may be longer than the predetermined time interval in step S22 (for example, several tens of minutes or more). However, if there are multiple automated voice response devices 1 connected to one order management server 5, it is necessary to maintain consistency between the databases of the automated voice response devices 1. Therefore, the second predetermined time interval must be at least the same as the first predetermined time interval, or shorter. In other words, the second predetermined time interval is set appropriately depending on the first predetermined time interval and the configuration of the automated voice response system.

[0105] As explained above, according to this embodiment, during voice interaction between S23 and S26, there is no access to the external order management server 5, and processing is performed by referring to the local database. Therefore, it is not affected by response delays from the external server, enabling smooth voice interaction. Furthermore, by performing the order processing to the external server asynchronously, even if the external server is undergoing maintenance or is under heavy load, the acceptance of orders from users can continue without interruption, thus improving the robustness of the system.

[0106] (Fourth Embodiment) In the third embodiment, the automated voice response system 1 received purchase information via voice calls. In this embodiment, it further has a function to respond to user inquiries via voice calls. This makes it possible to provide product explanations, recommendations based on purchase history, inquiries based on response history, payment processing, and cancellation processing when responding to user calls. Note that this embodiment has many components in common with the previous embodiment, so the differences will be explained in detail.

[0107] (System Configuration) In this embodiment, the order receiving system includes a purchasing information management server 4, as shown in Figure 7. The purchasing information management server 4 is configured to communicate with the voice automated response device 1 via a network NW. The purchasing information management server 4 is preferably configured separately from the voice automated response device 1, either as a standalone server or a server implemented in the cloud, from the standpoint of load balancing and fault avoidance, but it may also be configured integrally with the voice automated response device 1 or the order management server 5.

[0108] (Configuration of Purchasing Information Management Server 4) The purchasing information management server 4 is a server device that integrates sales management functions for managing the sales history of products and services, and customer relationship management (CRM) functions for managing user attributes and preferences. The purchasing information management server 4 has a "customer database" and a "product details database" in its storage unit. The purchasing information management server 4 has functions to manage payments, inquiries, and purchase history for orders that have been confirmed as orders by the order management server 5, in cooperation with the order management server 5. In addition, the purchasing information management server 4 also has functions to manage information on products and services sold.

[0109] The "Customer Database" uses user information such as the phone number of user terminal 2 as a key to store basic attributes of the user, such as their name, address, and email address, as well as past purchase history (purchase date and time, purchased items, amount), inquiry history, and user preference information analyzed by CRM functions. The "Product Details Database" manages detailed information about the products and services handled (name, price, specifications, detailed description) and information on recommended products related to each product (products for upselling and crossselling). Real-time inventory levels of products are managed by the order management server 5.

[0110] The automated voice response system 1 can use the API function of the purchasing information management server 4 to obtain purchasing-related data consisting of user attribute information, purchase history, inquiry history, user preference information, detailed information on products and services handled, information such as recommended products related to each product, and payment information for products and services, from user information such as telephone numbers and user inquiries obtained through voice calls.

[0111] (Configuration of the control unit 101 of the automated voice response device 1) In this embodiment, the functional configuration of the control unit 101 of the voice automatic response device 1 is further enhanced by the configuration of the third embodiment, including a purchase-related data acquisition unit 1017 and an inquiry handling unit 1018, as shown in Figure 8.

[0112] (Purchase-related data acquisition unit 1017) The purchase-related data acquisition unit 1017 requests purchase-related data from the purchase information management server 4 via the communication unit 103, using the API function, based on the user's telephone number obtained by the voice call unit 1013 and the user's inquiry content obtained through the voice call. The key information when requesting purchase-related data via API access includes the user's attribute information (name, address, etc.) included in the inquiry content, and information about the product and service being inquired about. The user's purchase-related data consists of data such as the user's attribute information, purchase history, inquiry history, user preference information, detailed information about the products and services handled, information such as recommended products related to each product, and payment information for products and services. The purchase-related data acquisition unit 1017 stores the acquired purchase-related data in the storage unit 102.

[0113] (Inquiry Support Department 1018) The inquiry handling unit 1018 retrieves purchase-related data from the storage unit 102 related to the user whose inquiry is being handled. The inquiry handling unit 1018 retrieves the call text from the voice call unit 1013 and sends the retrieved call text, purchase-related data, and a prompt instructing the inquiry handling to the large-scale language model via the communication unit 103. The large-scale language model is an external large-scale language model provider server connected by a network NW, and the data is sent via an API.

[0114] The prompts that the inquiry response unit 1018 sends to the large-scale language model include instructions such as: "1. Respond to the user's inquiry in a conversational format," "2. Generate the response by referring to the interaction history in the purchase-related data for tone, conversation content, etc.," "3. If the inquiry is about which product or service to choose, generate a response that includes recommendations by referring to the product information in the purchase-related data," and "4. If the inquiry concerns a procedure, refer to the purchase history and payment information in the purchase-related data to generate a response about the procedure, and if the user needs to perform the procedure, include in the response a statement that the information for that will be sent separately as a text message. Also output the information necessary to send the text message."

[0115] The inquiry handling unit 1018 receives the response text of the prompt sent from the large-scale language model, sends the response text to the voice call unit 1013, and the voice call unit 1013 converts the response text into speech and sends it to the user terminal 2, performing a voice call response. The inquiry handling unit 1018 continues the voice call until the inquiry handling with the user is completed, at which point the voice call is terminated. Although not shown in the flowchart of Figure 10, if the user wishes to purchase a product during the inquiry handling process, the process proceeds to step S24 to obtain the order request information and execute the order acceptance process. The flow of the order acceptance process is the same as in the third embodiment.

[0116] (Operation of the automated voice response system 1) The following describes each process of the automated voice response device 1 in this embodiment. Figure 10 is a flowchart of the process operation. The difference from the process flow in Figure 9 in the third embodiment is steps S31 and S32.

[0117] Referring to Figure 10, the flow of the order acceptance process performed by the automated voice response device 1 according to this embodiment will be described. In this embodiment, only steps S31 and S32 will be described. The processing of the other steps is the same as in the third embodiment and will therefore be omitted.

[0118] (Step S31) The purchase-related data acquisition unit 1017 requests purchase-related data from the purchase information management server 4 via the communication unit 103, using the API function, based on the user's telephone number obtained by the voice call unit 1013 and the user's inquiry content obtained through the voice call. The purchase-related data consists of data such as user attribute information, purchase history, inquiry history, user preference information, detailed information on products and services handled, information such as recommended products related to each product, and payment information for products and services. The purchase-related data acquisition unit 1017 stores the acquired purchase-related data in the storage unit 102.

[0119] (Step S32) The inquiry handling unit 1018 retrieves purchase-related data from the storage unit 102 related to the user whose inquiry is being handled. The inquiry handling unit 1018 retrieves the call text from the voice call unit 1013 and sends the retrieved call text, purchase-related data, and a prompt instructing the inquiry handling to the large-scale language model via the communication unit 103. The large-scale language model is an external large-scale language model provider server connected by a network NW, and the data is sent via an API.

[0120] The inquiry handling unit 1018 receives the response text of the prompt sent from the large-scale language model, sends the response text to the voice call unit 1013, and the voice call unit 1013 converts the response text into speech and sends it to the user terminal 2, performing a voice call response. The inquiry handling unit 1018 continues the voice call until the inquiry handling with the user is completed, at which point the voice call is terminated. Although not shown in the flowchart of Figure 10, if the user wishes to purchase a product during the inquiry handling process, the process proceeds to step S24 to obtain the order request information and execute the order acceptance process. The flow of the order acceptance process is the same as in the third embodiment.

[0121] As described above, the automated voice response device 1 according to the third embodiment is configured to asynchronously perform synchronization processing with the external order management server 5 (acquisition of inventory information) and voice dialogue processing with the user (provisional order). This enables highly responsive and smooth dialogue without causing waiting time to access the external server during voice dialogue. Furthermore, by performing the actual order processing to the external server retrospectively and periodically, high robustness is achieved, allowing the service to continue without losing order opportunities even in the event of temporary communication failures or high load on the external server.

[0122] Furthermore, the automated voice response system 1 according to the fourth embodiment is configured to further cooperate with the purchase information management server 4 to acquire detailed data such as user attribute information and past purchase history, and incorporate this into prompts for the large-scale language model. This enables not only routine order processing, but also product recommendations tailored to individual users' preferences and inquiry handling based on past history, thereby providing a high-quality customer experience comparable to that of a human operator.

[0123] (Fifth embodiment) In the first to fourth embodiments, a configuration was described in which either reservation acceptance or order acceptance is performed. In the fifth embodiment, a configuration is described in which a single automated voice response device 1 has both reservation acceptance and order acceptance functions, and switches processing according to the content of the conversation with the user.

[0124] (System Configuration) The system in this embodiment comprises a voice automated response device 1, a user terminal 2, a reservation management server 3, a purchase information management server 4, and an order management server 5, all connected via a network NW. Although not shown separately, the configuration is an integrated representation of Figures 4 and 7.

[0125] (Functional Configuration of the Automatic Voice Response System 1) The control unit 101 in this embodiment includes all the functional units described in the second and fourth embodiments. Specifically, it includes an information registration unit 1011, a reservation information acquisition unit 1012, an inventory information acquisition unit 1019, a voice call unit 1013, a provisional reservation processing unit 1014, a provisional order processing unit 1020, a final reservation processing unit 1015, a final order processing unit 1021, a notification unit 1016, a purchase-related data acquisition unit 1017, and an inquiry handling unit 1018.

[0126] In this embodiment, the information registration unit 1011 stores both access information to the reservation management server 3 and access information to the order management server 5 for a store that is both a service provider and a product seller. It also stores a telephone number for receiving calls from users.

[0127] The reservation information acquisition unit 1012 and the inventory information acquisition unit 1019 independently and asynchronously access the reservation management server 3 and the order management server 5 at predetermined time intervals, and keep the reservation acceptance database and the order acceptance database constantly updated to the latest state.

[0128] (Operation of the automated voice response system 1) A characteristic process in this embodiment is the dialogue control by the voice call unit 1013 and the inquiry handling unit 1018. The voice call unit 1013 sends prompts to the large language model that include the following instructions: "1. Identify through dialogue whether the purpose of the user's call is to make a service reservation, order a product, or make another inquiry," "2. If the purpose is a reservation, obtain reservation request information," "3. If the purpose is an order, obtain order request information."

[0129] The automated voice response system 1 performs the following processes according to the purpose identified through interaction with the user: (1) If the purpose is "reservation": The provisional reservation processing unit 1014 refers to the reservation acceptance database and performs provisional reservation processing (equivalent to step S05). (2) If the purpose is "order": The provisional order processing unit 1020 refers to the order acceptance database and performs provisional order processing (equivalent to step S25). (3) For other inquiries: The inquiry response unit 1018 generates a response based on purchase-related data and prompts obtained from the purchase information management server 4.

[0130] The reservation processing to the reservation management server 3 (equivalent to step S07) and the order processing to the order management server 5 (equivalent to step S27), which are performed in conjunction with the status updates of the reservation acceptance database and the order acceptance database, are each executed asynchronously and in parallel at independent predetermined time intervals.

[0131] As described above, according to the fifth embodiment, in business types such as restaurants where both seat reservations and takeout orders occur, a single telephone number and an automated voice response system 1 can automatically understand the user's intent and immediately access the appropriate database (reservation slots or inventory) to complete the order. This significantly reduces the telephone answering work for the store while providing users with a smooth, integrated service experience with no waiting time.

[0132] The present inventors have described the invention in detail based on embodiments above, but the present invention is not limited to the above embodiments. Furthermore, the above embodiments are described in detail for the purpose of explaining the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the described configurations. Some of the configurations of the above embodiments can be added, deleted, or replaced with other configurations.

[0133] Each of the above configurations, functions, and processing means may be implemented in hardware, in whole or in part, for example, by designing them as integrated circuits. Alternatively, each of the above configurations and functions may be implemented in software by having a processor interpret and execute programs that implement each function. Information such as programs, tables, and files that implement each function can be stored in memory, hard disks, SSDs, or other recording devices, or in recording media such as IC cards, SD cards, or DVDs. [Explanation of symbols]

[0134] 1. Automated voice response system, 2. User terminal, 3. Reservation management server, 4. Purchase information management server, 5. Order management server, 101. Control unit, 1011. Information registration unit, 1012. Reservation information acquisition unit, 1013. Voice call unit, 1014. Provisional reservation processing unit, 1015. Final reservation processing unit, 1016. Notification unit, 1017. Purchase-related data acquisition unit, 1018. Inquiry handling unit, 1019. Inventory information acquisition unit, 1020. Provisional order processing unit, 1021. Final order processing unit.

Claims

1. A voice automated response method performed by a computer connected via a network to a user terminal, a purchase information management server, and a reservation management server, The steps include: accepting a voice call request from the user terminal at a designated telephone number; The steps include obtaining user information from the user terminal, The steps include: Executing a voice interaction with a user based on purpose-specific prompts directed to a large-scale language model, and determining through the interaction whether the purpose of the user's call is to book a service or to make an inquiry; The steps include: selectively performing either (1) or (2) below to generate a response text, depending on the aforementioned objective; The steps include converting the generated response text into speech and sending it to the user terminal, Includes, The steps that are selectively performed are: (1) If the purpose is an inquiry, Based on the content of the inquiry and the user information, purchase-related data is obtained from the purchase information management server. Based on the purchase-related data, the content of the inquiry, and the response generation prompt, the large-scale language model is made to generate response text. (2) If the purpose is to make a reservation, The system obtains the user's reservation request information through the voice interaction, checks the availability of the reservation request information by referring to the reservation acceptance database in the memory unit, and if a reservation is possible, causes the large-scale language model to generate a response text indicating that the reservation has been accepted. The reservation confirmation corresponding to the reservation request information is registered in the reservation acceptance database. Access the aforementioned reservation management server and execute the reservation processing for the requested reservation information. A voice automated response method.

2. On user terminals, purchase information management servers, reservation management servers, and computers connected via the network, The steps include: accepting a voice call request from the user terminal at a designated telephone number; The steps include obtaining user information from the user terminal, The steps include: Executing a voice interaction with a user based on purpose-specific prompts directed to a large-scale language model, and determining through the interaction whether the purpose of the user's call is to book a service or to make an inquiry; The steps include: selectively performing either (1) or (2) below to generate a response text, depending on the aforementioned objective; The steps include converting the generated response text into speech and sending it to the user terminal, Make it run, In the step that is selectively performed, (1) If the purpose is an inquiry, Based on the content of the inquiry and the user information, purchase-related data is obtained from the purchase information management server. Based on the purchase-related data, the content of the inquiry, and the response generation prompt, the large-scale language model is made to generate response text. (2) If the purpose is to make a reservation, The system obtains the user's reservation request information through the voice interaction, checks the availability of the reservation request information by referring to the reservation acceptance database in the memory unit, and if a reservation is possible, causes the large-scale language model to generate a response text indicating that the reservation has been accepted. The reservation confirmation corresponding to the reservation request information is registered in the reservation acceptance database. Access the aforementioned reservation management server and have it execute the reservation processing for the requested reservation information. An automated voice response program.

3. An automated voice response device comprising a memory for storing the automated voice response program described in claim 2, and a processor for executing the automated voice response program.

4. A voice automated response method performed by a computer connected via a network to a user terminal, a purchasing information management server, and an order management server, The steps include: accepting a voice call request from the user terminal at a designated telephone number; The steps include obtaining user information from the user terminal, The steps include: Executing a voice dialogue with a user based on purpose-specific prompts directed to a large-scale language model, and determining through the dialogue whether the purpose of the user's call is to order a product or to make an inquiry; The steps include: selectively performing either (1) or (2) below to generate a response text, depending on the aforementioned objective; The steps include converting the generated response text into speech and sending it to the user terminal, Includes, The steps that are selectively performed are: (1) If the purpose is an inquiry, Based on the content of the inquiry and the user information, purchase-related data is obtained from the purchase information management server. Based on the purchase-related data, the content of the inquiry, and the response generation prompt, the large-scale language model is made to generate response text. (2) If the purpose is to place an order, The system obtains the user's order information through the voice interaction, checks the inventory status of the order information by referring to the order acceptance database in the memory unit, and if the order is available, causes the large-scale language model to generate a response text indicating that the order has been accepted. The completion of the order corresponding to the aforementioned order information is registered in the order acceptance database. Access the order management server and execute order processing for the order information. A voice automated response method.

5. Computers connected via the network to user terminals, purchasing information management servers, and order management servers, The steps include: accepting a voice call request from the user terminal at a designated telephone number; The steps include obtaining user information from the user terminal, The steps include: Executing a voice dialogue with a user based on purpose-specific prompts directed to a large-scale language model, and determining through the dialogue whether the purpose of the user's call is to order a product or to make an inquiry; The steps include: selectively performing either (1) or (2) below to generate a response text, depending on the aforementioned objective; The steps include converting the generated response text into speech and sending it to the user terminal, Make it run, In the step that is selectively performed, (1) If the purpose is an inquiry, Based on the content of the inquiry and the user information, purchase-related data is obtained from the purchase information management server. Based on the purchase-related data, the content of the inquiry, and the response generation prompt, the large-scale language model is made to generate response text. (2) If the purpose is to place an order, The system obtains the user's order information through the voice interaction, checks the inventory status of the order information by referring to the order acceptance database in the memory unit, and if the order is available, causes the large-scale language model to generate a response text indicating that the order has been accepted. The completion of the order corresponding to the aforementioned order information is registered in the order acceptance database. Access the aforementioned order management server and execute the order processing for the aforementioned order information. An automated voice response program.

6. An automated voice response device comprising a memory for storing the automated voice response program described in claim 5, and a processor for executing the automated voice response program.

Images