Information processing systems, information processing methods, and programs

Abstract

This technology provides a way to control, based on learning control conditions, whether or not to register confirmed response content as case information, where to register it, and the weight of its use when using it to generate future response texts. [Solution] The information processing system obtains a response draft generated based on the target input information, obtains the confirmed response content after the user confirms the response draft, and obtains learning control conditions related to the target input information, the confirmed response content, or the registered case information. Based on the learning control conditions, the information processing system determines learning policy information including whether or not the confirmed response content can be registered as case information, the registration destination of the case information, and the usage weight for future use of the case information. Furthermore, based on the learning policy information, the information processing system controls the process of registering the confirmed response content as case information and the process of generating a new response draft using the registered case information.

Description

Technical Field

[0001] The present invention relates to an information processing system, an information processing method, and a program.

Background Art

[0002] In recent years, target input information such as reviews, word-of-mouth, inquiries, messages, etc. regarding products and services has been continuously posted, and operators of stores, businesses, and various accounts are required to appropriately respond to this target input information. For this reason, various technologies for generating response texts for target input information and assisting response work have been proposed conventionally.

[0003] For example, in Patent Document 1, when there is a review comment on a review post for a company or store, it is determined whether the evaluation score attached to the post is lower than a predetermined standard or whether there are negative elements in the post, and according to the determination result, an information processing device that outputs a reply text prompt for requesting to generate a reply text for the review comment to a generation AI device is disclosed.

[0004] Also, in Patent Document 1, there are disclosed configurations for generating a reply text for conveying an apology, a declaration of improvement, or gratitude when there are negative elements, a configuration for generating a reply text based on an improvement plan or a user improvement plan, and a configuration for generating a reply text based on a corporate philosophy, concept, or store characteristics when the evaluation score is above the standard.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, when using the user's finalized response based on the generated response draft for subsequent response draft generation, simply registering the finalized response as example information and using it under the same conditions may not sufficiently improve the quality of response draft generation.

[0007] For example, while some confirmed responses may be suitable for future response drafting, others may be unsuitable for registration, should be registered in a specific case pool, or require weighting adjustments for future use.

[0008] In this regard, conventional technology still has room for improvement in that it determines whether or not to register the confirmed response obtained after a user confirmation operation, the registration destination, and learning policy information including the usage weight for future use, and controls the registration process, usage process, and generation process based on said learning policy information.

[0009] Therefore, the present invention aims to enable the control of whether or not to register the confirmed response content as case information, the registration destination, and the usage weight, based on learning control conditions, when using the confirmed response content for future response text generation. [Means for solving the problem]

[0010] According to one aspect of the present invention, an information processing system is provided. The information processing system has one or more control units, each of which acquires a response draft generated based on target input information, acquires a confirmed response content that has been confirmed by a user after a confirmation operation on the response draft, acquires learning control conditions related to the target input information, the confirmed response content, or registered case information, determines learning policy information including whether or not the confirmed response content can be registered as case information, the registration destination of the case information, and the usage weight for future use of the case information, and controls the process of registering the confirmed response content as case information and the process of generating a new response draft using the registered case information based on the learning policy information. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 shows an example of the system configuration of an information processing system. [Figure 2] Figure 2 shows an example of the hardware configuration of a user terminal. [Figure 3] Figure 3 shows an example of the hardware configuration of the management server. [Figure 4] Figure 4 shows an example of the functional configuration of a user terminal. [Figure 5] Figure 5 shows an example of the functional configuration of the management server. [Figure 6] Figure 6 is a flowchart showing an example of the process flow for controlling the registration and use of confirmed response content. [Figure 7] Figure 7 is a flowchart showing an example of the process flow for generating new response drafts using case information. [Modes for carrying out the invention]

[0012] The embodiments will be described below with reference to the drawings. The various features shown in the embodiments below can be combined with each other.

[0013] <Embodiment 1> 1. System Configuration Figure 1 shows an example of the system configuration of the information processing system 1000. As shown in Figure 1, the information processing system 1000 may include a user terminal 20 and a management server 30. Furthermore, the information processing system 1000 may include an external generation processing device 40 and a platform server 50 as needed.

[0014] The user terminal 20, management server 30, external generation processing device 40, and platform server 50 may be connected to each other via a network 60. The network 60 may be a wired or wireless communication network and may include, for example, a wide area network (WAN), a local area network (LAN), the internet, an enterprise network, a mobile communication network, or any combination thereof.

[0015] The user terminal 20 may be an information processing device used by a store, business operator, account manager, person in charge, or other user. The user terminal 20 may be, for example, a personal computer (PC), notebook PC, tablet terminal, smartphone, etc. In this embodiment, the user terminal 20 may have extension modules, add-ons, plugins, or dedicated applications that run on a web browser installed on it.

[0016] The user terminal 20 may acquire target input information displayed on an administration screen, message screen, inquiry screen, or other interface provided by the platform server 50. The target input information may be, for example, reviews, word-of-mouth, inquiries, messages, reservation requests, free-response answers in surveys, or other user input information. The user terminal 20 may display a response draft generated based on the target input information, accept confirmation, editing, approval, or other confirmation operations from the user, and acquire the confirmed response content after the confirmation operation.

[0017] The management server 30 may be a device that stores various data used in the information processing system 1000 and provides that data in response to requests from user terminals 20. The management server 30 may store, for example, user authentication information, contract plan information, store information, business information, account information, target input information, response drafts, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage history information, and other setting information.

[0018] The learning control conditions may be condition information related to the target input information and the confirmed response content. The learning control conditions may include, for example, evaluation information associated with the target input information, the content of the target input information, the content of the confirmed response content, the confirmation time, store information, operator information, account information, industry type information, platform type information, and other condition information used for the management of response text generation or case information.

[0019] The learning policy information may be information for managing the confirmed response content as case information for subsequent response text generation. The learning policy information may include a registration permission condition for determining whether the confirmed response content can be registered as case information, a pool designation condition for determining the registration destination of the case information, and a usage weight condition for determining the contribution degree when the case information is used in the future.

[0020] The management server 30 may manage a plurality of case pools. The case pools may be managed in association with, for example, stores, operators, accounts, industry types, service types, platform types, and other classifications. The management server 30 may determine the case pool for registering the confirmed response content based on the pool designation condition included in the learning policy information.

[0021] The user terminal 20 or the management server 30 may determine the learning policy information based on the learning control conditions. For example, the user terminal 20 or the management server 30 may determine whether to register the confirmed response content as case information, the case pool to be the registration destination if registration is performed, and the usage weight when the registered case information is used for future response text generation based on the learning control conditions related to the target input information and the confirmed response content.

[0022] The user terminal 20 or the management server 30 may control the registration process for registering confirmed response content as case information based on registration eligibility conditions and pool designation conditions. For example, confirmed response content determined to be eligible for registration based on registration eligibility conditions may be registered in the case pool designated by the pool designation conditions. On the other hand, confirmed response content determined not to be eligible for registration based on registration eligibility conditions may be controlled not to be registered as case information.

[0023] The user terminal 20 or the management server 30 may control the usage process for generating response drafts based on usage weight conditions. For example, registered case information may be referenced during response draft generation based on the contribution level determined by the usage weight conditions. This makes it possible to differentiate how case information registered in the same case pool is used in future response draft generation.

[0024] The external generation processing device 40 may be an external service that provides a generation model, language model, or inference application programming interface (API) for generating response text. The user terminal 20 or management server 30 may provide the external generation processing device 40 with target input information, selected and weighted case information based on learning policy information, and other input information to obtain a response text.

[0025] The platform server 50 may be an external system that provides an interface containing the target input information to the user terminal 20. The platform server 50 may be a server that provides, for example, an online review service, a store information service, a map service, a word-of-mouth management service, an inquiry management service, a reservation management service, a message management service, or other services. The user terminal 20 may display the screen provided by the platform server 50 on a web browser and perform operations such as acquiring the target input information, displaying a draft response, acquiring the confirmed response content, and other processes on that screen.

[0026] When generating a response draft corresponding to new target input information, the user terminal 20 or management server 30 may select case information to be used based on the learning policy information and weight the selected case information based on the usage weight conditions. Then, the user terminal 20, management server 30, or external generation processing device 40 may use the selected and weighted case information to generate a response draft corresponding to the new target input information.

[0027] In this embodiment, the generation of response drafts for target input information and the management of confirmed response content as example information are described as examples, but the processing targets are not limited to these. For example, the information processing system 1000 may apply similar processing to response tasks such as replying to reviews, answering inquiries, providing guidance for reservation requests, responding to messages, responding to free-response answers in questionnaires, and other response tasks.

[0028] The main processes described below are assumed to be performed by the user terminal 20, but are not limited to this. For example, at least a portion of the acquisition of target input information, acquisition of confirmed response content, acquisition of learning control conditions, determination of learning policy information, control of registration processing, control of usage processing, and control of generation processing may be performed by the management server 30 or the external generation processing device 40.

[0029] The information processing system described in the claims may consist of multiple devices or a single device. For example, some or all of the functions of the user terminal 20 and the management server 30 may be implemented within the same device, or the functions of the external generation processing device 40 may be incorporated into the management server 30 or the user terminal 20.

[0030] 2. Hardware Configuration (1) Hardware configuration of user terminal 20 Figure 2 shows an example of the hardware configuration of a user terminal 20. As shown in Figure 2, the user terminal 20 includes, as a hardware configuration, a control unit 210, a storage unit 220, an input unit 230, an output unit 240, a communication unit 250, and an internal bus 260. The control unit 210, the storage unit 220, the input unit 230, the output unit 240, and the communication unit 250 are electrically connected via the internal bus 260.

[0031] The control unit 210 is one of the following: a Central Processing Unit (CPU), a Micro Processing Unit (MPU), a System-on-a-Chip (SoC), a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), a Graphics Processing Unit (GPU), or a combination of at least two of these, and controls the entire user terminal 20 and executes processing in response to information input from, for example, the input unit 230, the communication unit 250, or the storage unit 220.

[0032] The storage unit 220 is one of the following: Read Only Memory (ROM), Random Access Memory (RAM), Flash Memory, Solid State Drive (SSD), or a combination of at least two of these, and stores the program and data used by the control unit 210 when it executes processing based on the program. The storage unit 220 may store programs corresponding to extension modules, add-ons, plugins, or dedicated applications that operate on a web browser, target input information, response drafts, confirmed response content, learning control conditions, learning policy information, case information, case pool information, display control information, operation history information, and other temporary processing data. The storage unit 220 is an example of a non-temporary recording medium that can be read by the computer on which the program is recorded.

[0033] In this specification, the data used by the control unit 210 when executing processing based on the program is described as being stored in the storage unit 220, but it may also be stored in the storage unit of another device that can communicate with the user terminal 20. The data may be stored in the storage unit of any device as long as the control unit 210 can access and retrieve it.

[0034] The control unit 210 executes processing based on the program stored in the memory unit 220, thereby realizing the processing shown in the functional configuration and flowcharts described later. For example, the control unit 210 can perform at least some of the following: acquiring target input information, displaying response drafts, receiving confirmation operations by the user, acquiring confirmed response content, acquiring learning control conditions, determining learning policy information, controlling the registration process to register confirmed response content as example information, controlling the utilization process to use registered example information for response draft generation, and controlling the generation process for response drafts corresponding to new target input information. In this embodiment, the main processing for response support and learning policy control is described as being performed on the user terminal 20, but is not limited to this.

[0035] The input unit 230 is a device that inputs information to the user terminal 20 in response to user operations. The input unit 230 receives operation inputs made by the user. The operation inputs are transmitted as command signals to the control unit 210 via the internal bus 260. The control unit 210 can perform predetermined controls and calculations based on the transmitted command signals as needed. The input unit 230 may be included in the casing of the user terminal 20 or it may be externally mounted. For example, the input unit 230 may be implemented as a touch panel integrated with the output unit 240. When the input unit 230 is implemented as a touch panel, the user can input tap operations, swipe operations, drag operations, etc. to the input unit 230. The input unit 230 may be a keyboard, mouse, trackpad, switch buttons, etc. instead of a touch panel.

[0036] The output unit 240 is a display unit, such as a display, that outputs information as a screen of a graphical user interface (GUI) that can be operated by the user. The output unit 240 may be included in the housing of the user terminal 20 or it may be an external device. More specifically, the output unit 240 may be implemented as a display device such as a liquid crystal display, an organic electroluminescent display, or a plasma display. The output unit 240 may display various information such as target input information, response drafts, confirmed response content, learning control conditions, learning policy information, case information, case pool information, user operation guidance information, and other information.

[0037] The communication unit 250 is a communication interface that connects the user terminal 20 to the network 60 and mediates communication with other devices. The communication unit 250 can be implemented, for example, by hardware that realizes wired and wireless communication, and circuits that control them. Based on instructions from the control unit 210, the communication unit 250 performs data transmission and reception processing via the network 60 according to predetermined communication protocols such as Transmission Control Protocol (TCP) / Internet Protocol (IP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), and HTTP over Transport Layer Security (HTTPS). For example, the communication unit 250 can send and receive authentication information, contract plan information, target input information, response draft, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage history information, and other information with the management server 30. Furthermore, depending on the embodiment, the communication unit 250 may send and receive response draft generation requests, generation results, screen information, input information, and other information with the external generation processing device 40 or platform server 50.

[0038] The hardware configuration of the user terminal 20 is not limited to the example described above. For example, the user terminal 20 may include multiple control units 210, or the storage unit 220 and communication unit 250 may be implemented in a distributed manner by multiple devices. Furthermore, the user terminal 20 is not limited to a personal computer, but may be a notebook computer, tablet terminal, smartphone, or other portable or stationary information terminal.

[0039] (2) Hardware configuration of the management server 30 Figure 3 shows an example of the hardware configuration of the management server 30. As shown in Figure 3, the management server 30 includes a control unit 310, a storage unit 320, a communication unit 330, and an internal bus 340 as its hardware configuration. The control unit 310, the storage unit 320, and the communication unit 330 are electrically connected via the internal bus 340.

[0040] The control unit 310 is one of the following: a Central Processing Unit (CPU), a Micro Processing Unit (MPU), a System-on-a-Chip (SoC), a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), a Graphics Processing Unit (GPU), or a combination of at least two of these, and controls the entire management server 30 and executes processing in response to information input from, for example, a user terminal 20, an external generation processing unit 40, a platform server 50, or a storage unit 320.

[0041] The storage unit 320 is one of the following: a hard disk drive (HDD), read-only memory, random access memory, solid-state drive, or a combination of at least two of these, and stores data used by the program and control unit 310 when executing processing based on the program. The storage unit 320 may store user authentication information, contract plan information, store information, business information, account information, target input information, response draft, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage weight information, registration history information, usage history information, and other setting information. Various types of data may be managed in relational databases, key-value stores, document databases, vector databases, or any other format.

[0042] The management server 30 may be configured as a device primarily responsible for memory processing, authentication processing, learning policy management processing, case information management processing, and data provision processing. For example, the control unit 310 may read and provide case information, case pool information, learning policy information, authentication-related information, and other information in response to a request from the user terminal 20, receive the confirmed response content and learning control conditions transmitted from the user terminal 20, and control the registration processing of confirmed response content and the processing of using registered case information based on the learning policy information. On the other hand, the main processing such as acquiring target input information, displaying response drafts, accepting confirmation operations by the user, and acquiring confirmed response content may be performed by the user terminal 20.

[0043] The communication unit 330 is a communication interface that connects the management server 30 to the network 60 and mediates communication with other devices. Based on instructions from the control unit 310, the communication unit 330 performs data transmission and reception processing via the network 60 according to a predetermined communication protocol. For example, the communication unit 330 can send and receive authentication information, contract plan information, target input information, response draft, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage history information, and other information with the user terminal 20. The communication unit 330 can also send and receive response draft generation requests and generation results with the external generation processing device 40.

[0044] The hardware configuration of the management server 30 is not limited to the example above. For example, the management server 30 may include multiple control units 310, and the storage unit 320 may be implemented as a single storage device or distributed across multiple storage devices. Similarly, the communication unit 330 may have multiple communication interfaces.

[0045] In this specification, the user terminal 20 performs the main processing, and the management server 30 primarily performs storage processing, authentication processing, learning policy management processing, case information management processing, and data provision processing. However, the entity executing the processing is not limited to this configuration. For example, the management server 30 may perform some of the processing that the user terminal 20 performs, or the user terminal 20 may perform some of the processing that the management server 30 performs. Furthermore, the external generation processing device 40 or the platform server 50 may perform some of these processing.

[0046] 3. Functional Configuration Figure 4 shows an example of the functional configuration of the user terminal 20. As shown in Figure 4, the user terminal 20 may include a local storage unit 410, a target input acquisition unit 420, a response draft acquisition unit 430, a confirmation operation reception unit 440, a confirmed response acquisition unit 450, a learning control condition acquisition unit 460, a learning policy determination unit 470, a registration control unit 480, a usage control unit 490, and a generation control unit 495. Furthermore, in one embodiment, the user terminal 20 may also include a display control unit 498.

[0047] Each of these functional units is a functional block realized by, for example, the control unit 210 executing a program stored in the memory unit 220 and controlling hardware such as the input unit 230, output unit 240, and communication unit 250. In this embodiment, the main processes such as acquiring target input information, acquiring response drafts, accepting confirmation operations, acquiring confirmed response content, acquiring learning control conditions, determining learning policy information, controlling registration processing, controlling usage processing, and controlling generation processing are described as being performed by the user terminal 20.

[0048] The local storage unit 410 may store target input information, response drafts, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage weight information, display control information, operation history information, and other information in progress. The local storage unit 410 may also store cache, setting information, generation request information, generation result information, and other information temporarily stored on the user terminal 20.

[0049] In this specification, "local storage unit 410" refers to the logical storage means in the functional configuration shown in Figure 4. The local storage unit 410 can be implemented in the storage unit 220 with the hardware configuration shown in Figure 2, or in an external storage device that can communicate with it. That is, the local storage unit 410 may be implemented on the same physical resources as the storage unit 220 as hardware, or it may be implemented distributed across multiple physical storage resources.

[0050] In this specification, "target input information" refers to input information that is the subject of response text generation. Target input information may include, for example, reviews, word-of-mouth, inquiries, messages, reservation requests, free-response answers in surveys, and other user input information.

[0051] In this specification, "response draft" refers to a draft text that shows the content of the response generated in response to the target input information. A response draft may include, for example, a reply draft to a review, a response draft to an inquiry, a guidance draft to a reservation request, a response to a message, and other draft texts.

[0052] In this specification, "confirmed response content" refers to the response content that has been confirmed by the user after a confirmation operation of the response draft. The confirmed response content may be content that the user has confirmed after editing the response draft, or content that the user has confirmed without editing the response draft.

[0053] In this specification, "learning control conditions" refer to conditional information used to control the use of registered case information for future response draft generation, after the confirmed response content has been registered as case information. Learning control conditions may include, for example, evaluation scores associated with target input information, the timing of confirmation of the confirmed response content, the content of the target input information, the content of the confirmed response content, legal risk scores, policy violation scores, store information, business information, account information, industry information, and platform type information.

[0054] In this specification, "learning policy information" refers to information for controlling the registration of confirmed response content as example information and the use of registered example information. Learning policy information may include registration eligibility conditions that determine whether or not confirmed response content can be registered as example information, pool designation conditions that determine the destination for registration of example information, and usage weight conditions that determine the contribution of example information when it is used in the future. In this specification, "registration eligibility conditions" refers to conditions for determining whether or not to register confirmed response content as example information. Registration eligibility conditions may include conditions for registering confirmed response content, conditions for not registering it, conditions for holding it, conditions for registering it after administrator confirmation, and other conditions that determine the registration status. In this specification, "pool designation conditions" refers to conditions for specifying the example pool to which confirmed response content will be registered when it is registered as example information. In this specification, "usage weight conditions" refers to conditions that determine the contribution of registered example information when it is used in future response draft generation. The weighting criteria may include weights for each case information, weights for each case pool, weights based on elapsed time, weights based on similarity to the target input information, or combinations thereof. Furthermore, the usage weight conditions are not limited to fixed weight values ​​stored in the memory unit, but may also be calculation conditions that determine the effective usage weight calculated at the time of use or acquisition of the case information, based on the base weight of the case information, the time elapsed since the registration or confirmation date, the similarity with the target input information, the evaluation range, the legal risk score, the policy violation score, or other conditions.

[0055] In this specification, "case information" refers to information that includes target input information and confirmed response content determined in response to said target input information, and which can be used for subsequent response text generation. Case information may be managed in association with classifications such as stores, businesses, accounts, industries, service types, platform types, and others. Furthermore, the industry, store category, and other classification information associated with case information may be classification information transmitted from the user terminal 20, classification information derived from contract information, store information, or account information stored in the management server 30, or common classification information obtained by mapping these.

[0056] In this specification, "case pool" means a memory area or logical collection for classifying and storing case information. A case pool may include, for example, case pools associated with stores, case pools associated with businesses, case pools associated with accounts, case pools associated with industries, case pools associated with service types, and other case pools. An industry-associated case pool is not limited to the industry name itself, but may be managed based on common industry identification information or pool identification information obtained by associating industry information belonging to multiple classification systems with industry mapping information.

[0057] The target input acquisition unit 420 acquires target input information. For example, the target input acquisition unit 420 may acquire target input information displayed on an administration screen, message screen, inquiry screen, or other interface provided by the platform server 50. The target input acquisition unit 420 may also acquire target input information selected by the user, target input information currently displayed, target input information that satisfies predetermined conditions, and other target input information.

[0058] The response text acquisition unit 430 acquires a response text generated based on the target input information. The response text acquisition unit 430 may acquire a response text generated within the user terminal 20, or it may acquire a response text from the management server 30 or an external generation processing device 40. The response text may be generated based on the target input information, case information, learning policy information, or other input information.

[0059] The confirmation operation reception unit 440 accepts confirmation operations from the user regarding the response draft. Confirmation operations may include, for example, operations to approve the response draft, operations to confirm the response draft after editing it, operations to confirm the response draft as a target for transmission, operations to confirm the response draft as a candidate for case registration, and other operations.

[0060] The confirmed response acquisition unit 450 acquires the confirmed response content after it has been confirmed through a confirmation operation received by the confirmation operation reception unit 440. For example, the confirmed response acquisition unit 450 may acquire the content entered in the reply input area, response input area, message input area, or other input areas based on the response draft as the confirmed response content.

[0061] The learning control condition acquisition unit 460 acquires learning control conditions related to the target input information, confirmed response content, or registered case information. For example, the learning control condition acquisition unit 460 may acquire evaluation scores associated with the target input information, the type of target input information, the content of the target input information, the content of the confirmed response content, the confirmation date, store information, business information, account information, industry information, platform type information, and other condition information.

[0062] The learning policy determination unit 470 determines learning policy information based on learning control conditions. For example, the learning policy determination unit 470 may determine, based on the learning control conditions, registration eligibility conditions that determine whether or not to register confirmed response content as example information, pool designation conditions that determine the registration destination when confirmed response content is registered as example information, and utilization weight conditions that determine the degree of contribution when using registered example information for future response text generation.

[0063] The registration control unit 480 controls the registration process for registering confirmed response content as case information based on registration eligibility conditions and pool designation conditions. For example, the registration control unit 480 may register confirmed response content that has been determined to be eligible for registration based on the registration eligibility conditions into the case pool designated by the pool designation conditions. On the other hand, the registration control unit 480 may suppress the registration of confirmed response content that has been determined not to be eligible for registration based on the registration eligibility conditions as case information.

[0064] The usage control unit 490 controls the usage process for generating response texts using registered case information based on usage weight conditions. For example, when generating a response text corresponding to new target input information, the usage control unit 490 may select case information to be used and assign weights to the selected case information based on usage weight conditions.

[0065] The generation control unit 495 controls the generation process that uses case information selected and weighted based on the learning policy information when generating a response draft corresponding to new target input information. For example, the generation control unit 495 may send a generation request including the target input information, selected case information, usage weights corresponding to the case information, and setting information as needed to the external generation processing unit 40, and obtain a response draft from the external generation processing unit 40.

[0066] The display control unit 498 controls the display of various information, including target input information, response drafts, confirmed response content, learning control conditions, learning policy information, case information, case pool information, and other information. For example, the display control unit 498 may display the response draft on a confirmation screen, pop-up, side panel, input candidate display area, or other display area. The display control unit 498 may also display whether or not the confirmed response content has been registered as case information, the case pool to which it will be registered, and information regarding usage weight for future use.

[0067] Figure 5 shows an example of the functional configuration of the management server 30. As shown in Figure 5, the management server 30 may include a server storage unit 510, an authentication management unit 520, a configuration information management unit 530, a learning policy management unit 540, a case pool management unit 550, a case information registration unit 560, a case information provision unit 570, and a generation request management unit 580.

[0068] Each of these functional units is a functional block realized, for example, by the control unit 310 executing a program stored in the storage unit 320 and controlling hardware such as the communication unit 330. In this embodiment, the management server 30 is described as being mainly responsible for storage processing, authentication processing, learning policy management processing, case information management processing, case pool management processing, and data provision processing.

[0069] The server storage unit 510 stores user authentication information, contract plan information, store information, business information, account information, target input information, response draft, confirmed response content, learning control conditions, learning policy information, case information, case pool information, usage weight information, registration history information, usage history information, and other information.

[0070] In this specification, "server storage unit 510" refers to the logical storage means in the functional configuration shown in Figure 5. The server storage unit 510 may be implemented in the storage unit 320 with the hardware configuration shown in Figure 3, or in an external storage device capable of communicating with it. The server storage unit 510 may be implemented on a single physical resource or distributed across multiple physical storage resources.

[0071] The authentication management unit 520 may perform user authentication based on authentication information transmitted from the user terminal 20. The configuration information management unit 530 may manage contract plan information, store information, business information, account information, industry information, industry mapping information, platform type information, learning policy setting information, and other configuration information provided to the user terminal 20. The industry mapping information may include information for associating the classification system of industry information transmitted from the user terminal 20 with the classification system of industry information derived from the contract information, store information, or account information of the authenticated account.

[0072] The learning policy management unit 540 manages learning policy information. For example, the learning policy management unit 540 may manage registration eligibility conditions, pool designation conditions, and usage weight conditions determined based on learning control conditions, associating them with store, business, account, or case information. The learning policy management unit 540 may also update learning policy information based on user settings or administrator settings.

[0073] The case pool management unit 550 manages multiple case pools. For example, the case pool management unit 550 may manage case pools in association with store, business operator, account, industry, service type, platform type, or other classifications. The case pool management unit 550 may also identify the case pool to which the confirmed response content will be registered based on pool designation conditions. Furthermore, the case pool management unit 550 may associate the first industry classification information transmitted from the user terminal 20 with the second industry classification information derived from the contract information, store information, or account information of the authenticated account, based on industry mapping information, with common industry identification information or pool identification information. This makes it possible to identify the second case pool corresponding to the same industry even if the method of identifying the industry used at registration and the method of identifying the industry used at acquisition are different.

[0074] The case information registration unit 560 processes the confirmed response content to be registered as case information. For example, the case information registration unit 560 may register the confirmed response content that has been selected for registration based on the registration eligibility conditions to the case pool specified by the pool designation conditions, in association with the target input information.

[0075] The case information provision unit 570 provides registered case information when generating a response draft corresponding to new target input information. For example, the case information provision unit 570 may provide case information selected and weighted based on usage weight conditions to the user terminal 20, the generation request management unit 580, or the external generation processing device 40. Furthermore, when providing registered case information, the case information provision unit 570 may dynamically calculate the effective utilization weight based on the base weight, registration date or confirmation date, evaluation range, similarity to target input information, and other information stored for the case information. In this case, the case information provision unit 570 may calculate the effective utilization weight based on the latest conditions at the time of acquiring the case information without updating the case information or utilization weight information stored in the memory unit, and provide case information weighted based on that effective utilization weight.

[0076] The generation request management unit 580 manages the generation requests for response drafts. For example, the generation request management unit 580 may generate a generation request that includes new target input information, selected and weighted case information based on learning policy information, and setting information as needed, and send it to the external generation processing unit 40. Alternatively, the generation request management unit 580 may obtain a response draft from the external generation processing unit 40 and provide it to the user terminal 20.

[0077] The functional units of the user terminal 20 and the management server 30 are not limited to the examples above. For example, the management server 30 may perform some of the functions that the user terminal 20 performs, or the user terminal 20 may perform some of the functions that the management server 30 performs. In addition, the external generation processing device 40 or the platform server 50 may perform some of these functions.

[0078] 4. Information Processing The information processing of Embodiment 1 will be described below.

[0079] (1) Overview of the process The information processing system 1000 obtains a response draft generated based on the target input information, obtains the confirmed response content after the user confirms the response draft, obtains learning control conditions related to the target input information, the confirmed response content, or registered case information, and determines learning policy information based on the learning control conditions. The learning policy information includes whether or not the confirmed response content can be registered as case information, the registration destination of the case information, and the usage weight for future use of the case information.

[0080] Furthermore, the information processing system 1000 controls the registration process of confirmed response content as example information and the generation process of new response drafts using the registered example information, based on the learning policy information. In one embodiment, the learning policy information may include registration eligibility conditions that determine whether or not confirmed response content can be registered as example information, pool designation conditions that determine the registration destination of the example information, and usage weight conditions that determine the contribution of the example information when it is used in the future. In this case, the information processing system 1000 may control the registration process to register confirmed response content as example information based on the registration eligibility conditions and the pool designation conditions, and may control the usage process that uses the registered example information for response draft generation based on the usage weight conditions. In addition, when generating a response draft corresponding to new target input information, the information processing system 1000 may control the generation process that uses example information selected and weighted based on the learning policy information.

[0081] Here, "target input information" refers to the input information that is the target of response text generation. Target input information may include, for example, reviews, word-of-mouth, inquiries, messages, reservation requests, free-response answers in surveys, and other user input information.

[0082] Furthermore, "obtaining the finalized response content" refers to obtaining the response content that was ultimately adopted after the user has confirmed, edited, approved, and sent the retrieved response draft. The finalized response content may be the content after the response draft has been edited, or it may be the content after the response draft has been adopted as is.

[0083] Furthermore, "acquiring learning control conditions" means acquiring conditional information for controlling the registration of confirmed response content as case information and the use of registered case information. Learning control conditions may include, for example, evaluation scores associated with target input information, content of target input information, content of confirmed response content, confirmation date, store information, business information, account information, industry information, platform type information, and other conditional information.

[0084] Furthermore, "determining learning policy information" means determining whether or not to register confirmed response content as example information, where to register it if it is registered as example information, and the weights to be used when using registered example information in future response draft generation, based on learning control conditions. Learning policy information may be information that directly indicates whether or not it can be registered, where it can be registered, and the weights to be used, or it may be information that is defined as registration eligibility conditions, pool designation conditions, and usage weight conditions.

[0085] Furthermore, "controlling the registration process" means controlling whether or not to register confirmed response content as case information, and if so, the case pool to which it will be registered, based on the learning policy information. "Controlling the generation process" means controlling the process of generating response drafts corresponding to new target input information using registered case information, based on the learning policy information. In one embodiment, "controlling the utilization process" means controlling to what extent registered case information is reflected in future response draft generation, based on the utilization weight conditions.

[0086] By performing this process, the information processing system 1000 does not uniformly register and use confirmed response content as case information, but rather determines learning policy information, including registration eligibility, registration destination, and usage weight, according to learning control conditions, and controls the registration of case information and the generation of new response drafts based on this learning policy information. This makes it easier to appropriately manage the case information used for generating response drafts and improves the quality of response draft generation corresponding to new target input information.

[0087] (2) Details of information processing The details of the information processing in this embodiment will be described below with reference to Figures 6 and 7. Figure 6 is a flowchart showing an example of a processing flow for controlling the registration and use of confirmed response content. Figure 7 is a flowchart showing an example of a processing flow for generating a new response draft using case information.

[0088] In this specification, "evaluation score" refers to information indicating the evaluation associated with the target input information. The evaluation score may be, for example, a number of stars, points, a rating scale, a number of icons, or other evaluation values. Furthermore, "evaluation range" refers to a division of the evaluation score into multiple ranges. The evaluation range may include, for example, a low-evaluation range, a medium-evaluation range, and a high-evaluation range.

[0089] In this specification, "elapsed time" refers to information indicating the time elapsed since the registration of case information, the finalization of the response content, or the generation of the response draft. Elapsed time may be expressed in days, months, years, whether a predetermined period has elapsed, or as a time index based on these.

[0090] In this specification, "Legal Risk Score" refers to an indicator that shows the potential for expressions contained in the confirmed response to be problematic from a legal standpoint, including laws, advertising regulations, industry regulations, rights infringement, defamation, personal information protection, and other legal perspectives. Furthermore, "Policy Violation Score" refers to an indicator that shows the potential for expressions contained in the confirmed response to be inconsistent with platform policies, business policies, store operation rules, prohibited word rules, and other operational regulations.

[0091] In this specification, "learning suitability score" refers to an index indicating the suitability of registering a finalized response as case information and using it for future response draft generation. The learning suitability score may be calculated based on evaluation scores, elapsed time, legal risk scores, policy violation scores, difference amounts, and other learning control conditions.

[0092] In this specification, "difference amount" refers to information indicating the difference between the draft response and the final response. The difference amount may be calculated based on, for example, the difference in character count, edit distance, amount of added text, amount of deleted text, amount of replacement, or other edits.

[0093] (Step S601) The target input acquisition unit 420 acquires target input information. The target input information may be acquired, for example, from an administration screen, message screen, inquiry screen, or other interface provided by the platform server 50. The target input acquisition unit 420 then proceeds to step S602.

[0094] (Step S602) The response draft acquisition unit 430 acquires a response draft generated based on the target input information. The response draft may be generated within the user terminal 20, or it may be acquired from the management server 30 or an external generation processing device 40. The response text acquisition unit 430 then proceeds to step S603.

[0095] (Step S603) The confirmation operation reception unit 440 accepts confirmation operations from the user regarding the response draft. Confirmation operations may include, for example, an operation to approve the response draft, an operation to confirm the response draft after editing it, an operation to confirm the response draft as a target for transmission, or other operations. The confirmation operation reception unit 440 then proceeds to step S604.

[0096] (Step S604) The confirmed response acquisition unit 450 acquires the confirmed response content after the confirmation operation in step S603. The confirmed response content may be the content after the response draft has been edited by the user, or it may be the content that has been confirmed without the response draft being edited. The confirmed response acquisition unit 450 then proceeds to step S605.

[0097] (Step S605) The learning control condition acquisition unit 460 acquires learning control conditions related to the target input information, confirmed response content, or registered case information. The learning control conditions may include, for example, an evaluation score associated with the target input information, the content of the target input information, the content of the confirmed response content, the confirmation date, store information, business information, account information, industry information, and platform type information. The learning control condition acquisition unit 460 then proceeds to step S606.

[0098] (Step S606) The learning policy determination unit 470 may identify the evaluation tier to which the evaluation score included in the learning control conditions belongs. For example, the learning policy determination unit 470 may classify the evaluation scores into low evaluation tier, medium evaluation tier, and high evaluation tier, and determine at least one of the registration eligibility condition, pool designation condition, and usage weight condition for each evaluation tier. As an example, if the evaluation score is expressed in terms of stars, the learning policy determination unit 470 may classify evaluation scores from 1 to 2 stars as the low evaluation tier, evaluation scores from 3 stars as the medium evaluation tier, and evaluation scores from 4 to 5 stars as the high evaluation tier. The learning policy determination unit 470 may, when generating response drafts for low evaluations or complaints, extract case information corresponding to that evaluation range as a search target or set a relatively high usage weight for such case information if the evaluation score belongs to the low evaluation range. The learning policy determination unit 470 may, when generating response drafts for normal responses or improvement promotions, extract case information corresponding to that evaluation range as a search target or set a relatively high usage weight for such case information if the evaluation score belongs to the high evaluation range. In this case, it is not necessarily required to create a different case pool for each evaluation tier. For case information registered in the same case pool, the evaluation tier may be retained as attribute information and used for search filtering, ranking, or calculation of usage weights when generating response drafts. In another embodiment, the learning policy determination unit 470 may designate a case pool for complaint handling, improvement guidance, or low-rated responses as the registration destination if the evaluation score belongs to the low evaluation range, a case pool for normal responses or improvement promotion as the registration destination if the evaluation score belongs to the medium evaluation range, and a case pool for positive responses as the registration destination if the evaluation score belongs to the high evaluation range. This makes it possible to treat the confirmed response content as a learning target differently depending on the evaluation range. The learning policy determination unit 470 then proceeds to step S607.

[0099] (Step S607) The learning policy determination unit 470 may obtain the elapsed time from the registration or confirmation date of registered case information and determine usage weight conditions that decrease the usage weight of the case information as the elapsed time increases. For example, the learning policy determination unit 470 may set a first usage weight for case information registered within 3 months, a second usage weight lower than the first usage weight for case information registered more than 3 months but within 1 year, and a third usage weight lower than the second usage weight for case information registered more than 1 year ago. The learning policy determination unit 470 may also continuously decrease the usage weight in a linear, exponential, logarithmic, or other manner according to the number of days elapsed since the registration or confirmation date. Furthermore, the learning policy determination unit 470 or the utilization control unit 490 may calculate the effective utilization weight on demand based on the base weight, elapsed time, evaluation range, similarity with the target input information, and other information when generating a response text or acquiring case information, rather than updating the utilization weight information stored in the memory unit each time for registered case information. In this case, the information processing system 1000 can use the registered case information to generate a response text using the effective utilization weight that reflects the latest elapsed time at the time of use, without rewriting the case information or utilization weight information in the memory unit. The learning policy determination unit 470 then proceeds to step S608.

[0100] (Step S608) The learning policy determination unit 470 may acquire a legal risk score and a policy violation score related to the confirmed response content. For example, the learning policy determination unit 470 may determine whether to register the confirmed response content as case information if the legal risk score is above a predetermined threshold. Alternatively, the learning policy determination unit 470 may calculate a policy violation score based on whether the confirmed response content contains prohibited words, exaggerated expressions, expressions containing personal information, expressions that may constitute defamation, or expressions that do not conform to the platform policy, and determine whether to register the confirmed response content as case information if the policy violation score is above a predetermined threshold. This makes it possible to suppress the accumulation of confirmed response content that may have legal risks or policy violations as case information used for future response draft generation. The learning policy determination unit 470 then proceeds to step S609.

[0101] (Step S609) The learning policy determination unit 470 may calculate a learning suitability score based on the evaluation score, elapsed time, legal risk score, and policy violation score. For example, the learning policy determination unit 470 may increase the learning suitability score as the evaluation score increases, and decrease the learning suitability score as the legal risk score or policy violation score increases. The learning policy determination unit 470 may also decrease the learning suitability score as the elapsed time from the registration time increases. As an example, the learning policy determination unit 470 may register the confirmed response content as case information if the learning suitability score is above the first threshold, register it in a pending state if it is above the second threshold but below the first threshold, and suppress registration if it is below the second threshold. Furthermore, the learning policy determination unit 470 may set the utilization weight higher as the learning suitability score increases, and set the utilization weight lower as the learning suitability score decreases. The learning policy determination unit 470 then proceeds to step S610.

[0102] (Step S610) The learning policy determination unit 470 may calculate the difference between the draft response and the finalized response. For example, the learning policy determination unit 470 may calculate the difference based on the difference in character count, edit distance, added text, deleted text, replaced text, or other edits. The learning policy determination unit 470 may then use the calculated difference, along with evaluation score, elapsed time, legal risk score, policy violation score, and other learning control conditions, as one of several determination elements for determining registration eligibility and usage weight conditions. The learning policy determination unit 470 then proceeds to step S611.

[0103] (Step S611) The learning policy determination unit 470 determines the pool designation conditions for which the confirmed response content will be registered. For example, the learning policy determination unit 470 may determine the case pool to which the confirmed response content will be registered from among multiple case pools, including at least a first case pool associated with stores and a second case pool associated with industries. If a third case pool associated with service types is provided, the learning policy determination unit 470 may determine the case pool to which the confirmed response content will be registered from among the first case pool, the second case pool, and the third case pool. The learning policy determination unit 470 may also determine the case pool to which the confirmed response content will be registered based on store category, business category, account category, platform type, evaluation range, risk category, or policy category. As an example, the learning policy determination unit 470 may specify the registration destination for the confirmed response content from among case pools divided by store category or industry category, such as food and beverage, beauty, medical, accommodation, education, etc. Furthermore, during the registration process, the learning policy determination unit 470 or the case pool management unit 550 may acquire the industry information transmitted from the user terminal 20 as first industry classification information and convert the first industry classification information into common industry identification information or pool identification information for the second case pool based on the industry mapping information. In this case, the learning policy determination unit 470 may identify the second case pool corresponding to the converted common industry identification information or pool identification information as the registration destination for the confirmed response content. The learning policy determination unit 470 then proceeds to step S612.

[0104] (Step S612) The registration control unit 480 controls the registration process for registering confirmed response content as case information based on registration eligibility conditions and pool designation conditions. For example, the registration control unit 480 may register confirmed response content determined to be eligible for registration based on the registration eligibility conditions in the case pool designated by the pool designation conditions, in association with the target input information. On the other hand, the registration control unit 480 may suppress the registration of confirmed response content determined not to be eligible for registration based on the registration eligibility conditions. The registration control unit 480 then proceeds to step S613.

[0105] (Step S613) The usage control unit 490 controls the usage process for using registered case information to generate response texts, based on usage weight conditions. For example, the usage control unit 490 may set a usage weight for each case piece of information and use case information with a higher usage weight as a relative priority for generating response texts. The utilization control unit 490 then proceeds to step S614.

[0106] (Step S614) The display control unit 498 may display whether the confirmed response content can be registered, the case pool to which it will be registered, the usage weight, and the content of the learning policy information. The display control unit 498 may also display the history of registration processing or usage processing in a manner that can be confirmed by the user.

[0107] Next, we will explain the process of generating new response drafts using case information. Figure 7 is an example of a flowchart for this process.

[0108] (Step S651) The target input acquisition unit 420 acquires new target input information. The new target input information may be, for example, newly posted reviews, word-of-mouth, inquiries, messages, reservation requests, free-response comments on surveys, or other input information. The target input acquisition unit 420 then proceeds to step S652.

[0109] (Step S652) The usage control unit 490 identifies a pool of cases associated with the new target input information. For example, the usage control unit 490 may identify a first case pool associated with a store and a second case pool associated with an industry, based on the store and industry corresponding to the new target input information. Furthermore, if a third case pool associated with a service type is provided, the usage control unit 490 may further identify the third case pool based on the service type corresponding to the new target input information. Furthermore, during the usage process, the usage control unit 490 or the case pool management unit 550 may derive industry information as second industry classification information based on the contract information, store information, or account information of the authenticated account, and convert the second industry classification information into common industry identification information or pool identification information of the second case pool based on the industry mapping information. In this case, the second industry classification information may belong to a different classification system than the first industry classification information transmitted from the user terminal 20 when the confirmed response content is registered. The usage control unit 490 may also obtain case information from the same second case pool by referring to the second case pool corresponding to the converted common industry identification information or pool identification information, even if the method of identifying the industry differs between registration and acquisition. The user control unit 490 then proceeds to step S653.

[0110] (Step S653) The usage control unit 490 may determine the mixing ratio of case information obtained from multiple identified case pools based on learning policy information. For example, the usage control unit 490 may determine the mixing ratio of case information obtained from a first case pool associated with stores and case information obtained from a second case pool associated with industries. If a third case pool associated with service types is provided, the usage control unit 490 may determine the mixing ratio of case information obtained from the first, second, and third case pools. The usage control unit 490 may also determine the mixing ratio of case information obtained from multiple case pools based on the store category, industry, service type, evaluation range, risk category, or platform type corresponding to the new target input information. As an example, if the new target input information concerns an existing store and sufficient case information has been accumulated for that store, the usage control unit 490 may set the ratio of case information obtained from the first case pool higher than the ratio of case information obtained from the second case pool. Furthermore, if sufficient case information has not been accumulated at the target store, the usage control unit 490 may set a higher ratio for case information acquired from the second case pool, or from the third case pool if a third case pool is provided. The utilization control unit 490 then proceeds to step S654.

[0111] (Step S654) The utilization control unit 490 may select case information to be used for response draft generation based on utilization weight conditions and mixing ratios, and may weight the selected case information. For example, the utilization control unit 490 may prioritize the selection of case information that is highly relevant to new target input information and has a high utilization weight. The utilization control unit 490 may also determine the contribution of each case information in response draft generation by combining the utilization weight for each case information and the mixing ratio for each case pool. Furthermore, the utilization control unit 490 may exclude case information with a high legal risk score or policy violation score from selection during response draft generation, or reduce its utilization weight. The user control unit 490 then proceeds to step S655.

[0112] (Step S655) The generation control unit 495 controls the response text generation process using new target input information, selected and weighted case information, and setting information as needed. For example, the generation control unit 495 may send a generation request containing this information to an external generation processing unit 40 and obtain a response text from the external generation processing unit 40. The generation control unit 495 then proceeds to step S656.

[0113] (Step S656) The display control unit 498 displays the generated response draft. The display control unit 498 may also display the type of case information used to generate the response draft, the case pool, the usage weight, and other information.

[0114] In one embodiment, the configuration information management unit 530 or the learning policy management unit 540 may accept user settings regarding whether or not to register confirmed response content, the registration destination, and the usage weight. The learning policy determination unit 470 or the learning policy management unit 540 may update the learning policy information based on the user settings. This makes it possible to adjust the registration and usage control of confirmed response content according to the operational policy of each store, business operator, or account.

[0115] In one embodiment, the generation control unit 495 may determine a processing mode from among a fully automatic mode that executes from the presentation of a response draft to transmission, an automatic mode with confirmation that executes transmission after user approval, and a proposal mode that only presents a response draft, based on the learning control conditions. For example, if the legal risk score or policy violation score is high, the generation control unit 495 may select a processing mode that indicates a high degree of user involvement.

[0116] Thus, according to this embodiment, when registering confirmed response content as case information, registration eligibility, registration destination, and usage weight can be controlled based on evaluation range, elapsed time, legal risk score, policy violation score, learning fit score, and difference amount. Therefore, compared to uniformly registering and using confirmed response content, it becomes easier to appropriately manage case information used for future response draft generation.

[0117] Furthermore, according to this embodiment, it is possible to determine the mixing ratio of case information obtained from multiple case pools, select and weight case information based on usage weight conditions, and then generate a response draft. Therefore, it becomes easier to use store-specific cases and industry-common cases in an appropriate ratio depending on the new target input information. In addition, if case pools associated with service types or other additional classifications are provided, it becomes easier to use cases from those additional case pools in an appropriate ratio as well.

[0118] Note that the execution order of the steps shown in Figures 6 and 7 is not limited to the examples above. For example, the processes from step S606 to step S611 may be executed in a different order, or at least some of them may be executed in parallel. Also, at least some of the processes shown in Figures 6 and 7 may be shared among the user terminal 20, the management server 30, the external generation processing device 40, and the platform server 50.

[0119] (3) Effects of the embodiment According to this embodiment, after a user confirms the response draft generated based on the target input information, the confirmed response content is obtained, and based on the learning control conditions related to the target input information and the confirmed response content, it is possible to determine whether the confirmed response content can be registered as example information, where the example information will be registered, and learning policy information including the usage weight for future use of the example information. This makes it easier to appropriately manage the example information used for future response draft generation compared to registering and using all confirmed response content as example information.

[0120] Furthermore, according to this embodiment, the process of registering confirmed response content as example information and the process of generating new response drafts using the registered example information can be controlled based on the learning policy information. Therefore, whether confirmed response content is to be registered, which example pool it is registered in, and to what extent it is reflected in future response draft generation can be adjusted according to the learning control conditions.

[0121] Furthermore, according to this embodiment, in one aspect, the learning policy information includes registration eligibility conditions, pool designation conditions, and usage weight conditions. Therefore, the registration process of confirmed response content can be controlled based on the registration eligibility conditions and pool designation conditions, and the usage process of registered case information can be controlled based on the usage weight conditions. In addition, when generating a response draft corresponding to new target input information, the generation process can be controlled to use case information selected and weighted based on the learning policy information. This makes it easier to use case information corresponding to the target input information and improves the quality of response draft generation.

[0122] Furthermore, according to this embodiment, registration eligibility conditions, pool designation conditions, and usage weight conditions can be determined based on the evaluation tier to which the evaluation score belongs. This makes it easier to register and utilize case examples according to the characteristics of each type of input information, such as low evaluation tier, medium evaluation tier, and high evaluation tier, which have different evaluation characteristics.

[0123] Furthermore, according to this embodiment, the usage weight of case information can be reduced according to the elapsed time since the registration of the case information. This suppresses the excessive influence of old case information on future response draft generation, making it easier to generate response drafts that are in line with relatively new operational policies.

[0124] Furthermore, according to this embodiment, it is possible to determine the registration criteria for whether or not to register a confirmed response as case information based on the legal risk score and the policy violation score. This makes it possible to suppress the accumulation of confirmed response content that may have legal risks or policy violations as case information used for future response draft generation.

[0125] Furthermore, according to this embodiment, a learning suitability score can be calculated based on the evaluation score, elapsed time, legal risk score, and policy violation score, and registration eligibility conditions and usage weight conditions can be determined based on the learning suitability score. This makes it possible to control the registration and use of confirmed response content by comprehensively considering multiple learning control conditions.

[0126] Furthermore, according to this embodiment, the difference between the draft response and the finalized response can be used as one of several determination elements for determining registration eligibility and usage weight conditions, along with evaluation score, elapsed time, legal risk score, policy violation score, and other learning control conditions. This makes it possible to consider the extent of modifications made by the user to the draft response in combination with other learning control conditions, thereby making it easier to more appropriately determine the usefulness of the finalized response as case information.

[0127] Furthermore, according to this embodiment, case information can be managed by dividing it into multiple case pools, including at least a first case pool associated with stores and a second case pool associated with industries, and the registration destination can be determined based on learning control conditions. This makes it easier to manage store-specific response trends and industry-common response trends separately. In addition, if case pools associated with service types or other additional classifications are further established, it becomes easier to manage response trends corresponding to those additional classifications separately.

[0128] Furthermore, according to this embodiment, when generating a response draft corresponding to new target input information, the mixing ratio of case information obtained from multiple case pools can be determined based on the learning policy information. This makes it easier to use store-specific cases and industry-common cases in an appropriate ratio according to the new target input information. In addition, if case pools associated with service types or other additional classifications are provided, it also becomes easier to use cases from those additional case pools in an appropriate ratio.

[0129] Furthermore, according to this embodiment, user settings regarding whether or not to register confirmed response content, the registration destination, and usage weight can be accepted, and learning policy information can be updated based on user settings. This makes it possible to adjust the control content of case information registration and usage according to the operational policies of each store, business operator, and account.

[0130] Furthermore, according to this embodiment, the processing mode can be determined from among the fully automatic mode, the automatic mode with confirmation, and the suggestion mode based on the learning control conditions. This makes it possible to appropriately control the processing after the response text is generated while adjusting the degree of user involvement according to the target input information and the confirmed response content.

[0131] (modified version) Hereinafter, Modifications 1 to 12 are described as examples of how the configuration and processing according to the embodiment can be implemented with various modifications. The configurations or processing described in each of the following modifications can be implemented in any combination as long as they do not contradict each other. Furthermore, the configurations or processing described in each modification are arbitrary configurations that can be additionally applied to the basic processing which, based on learning control conditions, obtains the confirmed response content that has been confirmed through a confirmation operation by the user for the response draft generated based on the target input information, determines learning policy information including whether or not the confirmed response content can be registered as example information, the registration destination of the example information, and the usage weight for future use of the example information, and controls the registration processing and generation processing based on said learning policy information. The basic processing can be executed even if the modifications are not applied.

[0132] (Modification 1: Division of responsibilities for executing the process) The division of processing among the user terminal 20, management server 30, external generation processing device 40, and platform server 50 is not limited to the above-described configuration. For example, the user terminal 20 may perform the acquisition of target input information, display of response drafts, and acquisition of confirmed response content, while the management server 30 may perform the acquisition of learning control conditions, determination of learning policy information, registration processing, and utilization processing. Alternatively, the user terminal 20 may determine at least a portion of the learning policy information, and the external generation processing device 40 may perform at least a portion of the selection or weighting of case information in addition to the response draft generation processing.

[0133] This configuration allows for flexible adjustment of processing load depending on client-side computing resources, server-side computing resources, communication load, response speed, security requirements, or the usage environment.

[0134] (Example 2: Transformation of target input information) The target input information is not limited to reviews, testimonials, inquiries, messages, reservation requests, and free-response questionnaires. For example, the target input information may also include chat logs, email bodies, transcripts of phone calls, customer requests, complaint records, post-visit questionnaires, sales meeting notes, support tickets, internal application documents, and other input information that can be used to generate response drafts.

[0135] With this configuration, the learning policy control of this embodiment can be applied not only to review response tasks, but also to various response tasks such as customer service, reservation handling, inquiry handling, sales support, and support handling.

[0136] (Variation 3: Modification of confirmation operation and confirmation response content) The confirmation operation is not limited to the operation of approving the response text. For example, the confirmation operation may be the operation of saving the response text after editing it, the operation of designating the response text as a recipient for transmission, the operation of confirming the response text as a draft, the operation of selecting the response text as a candidate for case registration, or the operation of linking the response text to an external system. Furthermore, the confirmed response content may be the response content that was actually sent, the response content that was saved as a draft before transmission, or the response content that the user designated as a learning target.

[0137] This configuration allows for flexible configuration of the timing and target of acquiring confirmed response content, according to the transmission specifications, user confirmation flow, and operational rules for each platform.

[0138] (Modification 4: Modification of learning control conditions) Learning control conditions are not limited to evaluation scores, elapsed time, legal risk scores, and policy violation scores. For example, learning control conditions may include the category of the target input information, sentiment judgment results, importance, urgency, poster attributes, language, region, store attributes, store category, business attributes, account attributes, industry, service type, platform type, source of response draft, amount of user editing, transmission results, customer reactions, whether or not a follow-up inquiry occurred, whether or not a complaint was filed, and other conditional information. The information processing system 1000 may also determine learning policy information or a processing mode based on the analysis results regarding the target input information and the learning control conditions or business control conditions. The analysis results may include candidate text, evaluation candidates, input area candidates, category candidates, sentiment judgment results, and other candidate information. Business control conditions may include evaluation ranges, risk scores, policy violation scores, progress, user permissions, platform specifications, and other conditional information.

[0139] This configuration allows for precise control over the registration and use of case information used in generating response drafts, depending on the content of the target input information, analysis results, operational status, and business risks. Furthermore, since it can be configured as a control loop using analysis results such as text candidates, evaluation candidates, and input area candidates, along with control conditions such as evaluation ranges, risk scores, and progress, it becomes easier to achieve learning policy control that is less dependent on specific generative models or features.

[0140] (Variation 5: Variation of the method for determining learning policy information) The method for determining learning policy information is not limited to a predetermined rule-based approach. For example, learning policy information may be determined based on rule-based processing, scoring processing, statistical processing, machine learning models, generative models, classification models, user settings, administrator settings, store-specific operational settings, or a combination thereof. Furthermore, learning policy information may be fixed, or it may be updated based on the accumulation of case information, usage history, user modification history, responses after a response, and other feedback.

[0141] With this configuration, learning policy information can be determined based on simple rules at the start of operation, and as operation continues, the learning policy information can be updated according to actual results.

[0142] (Variation 6: Variation of registration eligibility conditions) The registration eligibility criteria are not limited to binary decisions on whether or not to register the confirmed response. For example, registration eligibility criteria may include conditions for normal registration of the confirmed response, registration in a pending state, registration after administrator confirmation, temporary registration, storage while excluded from use, and reassessment after a certain period of time. Furthermore, registration eligibility criteria may be determined based on a combination of conditions such as legal risk score, policy violation score, difference amount, evaluation range, user settings, detection results of prohibited expressions, and other factors.

[0143] This configuration allows for flexible quality control of case information, as it enables managing confirmed response content not only by simply registering or discarding it, but also by managing it in states such as awaiting confirmation, on hold, temporarily stored, and suspended.

[0144] (Variation 7: Variations of the example pool and pool designation conditions) Case pools are not limited to those associated with stores, industries, and service types. For example, case pools may be managed by associating them with businesses, brands, accounts, regions, languages, rating ranges, customer attributes, response categories, store categories, platform types, risk categories, policy categories, campaign units, time periods, and other classifications. Furthermore, a single case information may be associated with multiple case pools, and primary registration locations and secondary reference locations may be managed separately. In addition, case pools may hierarchically include store-specific case pools, case pools common to store categories, case pools common to industries, and case pools common to the entire organization. It should be noted that all of the above case pools do not need to be established simultaneously. For example, the information processing system 1000 manages case information using a first case pool associated with stores and a second case pool associated with industries, and a third case pool associated with service types may be additionally established depending on the operational form, service expansion, or future implementation. Furthermore, even if a third case pool is not established, service type information may be retained as attribute information of the case information and used for search filtering, ranking, or calculation of usage weights. Furthermore, the industries corresponding to the second case pool do not need to be identified based on the same source or classification system at the time of registration and acquisition. For example, in the registration process, industry information such as food and beverage, beauty, medical, accommodation, etc., transmitted from the user terminal 20 may be acquired as first industry classification information, and this first industry classification information may be attached to the case information for management. On the other hand, in the acquisition process, the management server 30 may derive industry information as second industry classification information from the contract information, store information, or account information of the authenticated account. Even if the first industry classification information and the second industry classification information belong to different classification systems, the information processing system 1000 may control registration to the same second case pool and acquisition from the same second case pool by converting the first industry classification information and the second industry classification information into common industry identification information or pool identification information of the second case pool based on industry mapping information.

[0145] This configuration allows for the separate management of response trends specific to each store, each store category, each industry, and responses based on regional differences, language differences, platform differences, and risk classifications. Furthermore, by managing case information in multiple tiers of case pools, it becomes easier to combine store-specific case information with case information common to store categories or industries when applying new target input information. Moreover, even if the industry information transmitted from the user terminal 20 during registration and the industry information derived from the contract information of the authenticated account during acquisition belong to different classification systems, the same second case pool can be matched based on the industry mapping information. This allows for the appropriate management of industry-common case information while absorbing differences in processing entities, information sources, or classification systems during registration and use.

[0146] (Variation 8: Variation of the weighting conditions and mixing ratio) The usage weighting criteria are not limited to assigning a single weight to each case information. For example, the usage weighting criteria may be determined based on information such as the similarity of the case information, registration date, evaluation range, difference amount, number of uses, recent adoption rate, user modification rate, legal risk score, policy violation score, case pool type, and other information. Furthermore, when obtaining case information from multiple case pools, the mixing ratio for each case pool may be changed according to the type of target input information, store attributes, industry, service type, user settings, usage history, and other conditions. Furthermore, the mixing ratio of case information obtained from multiple case pools can be determined based on two or more existing case pools. For example, if a first case pool and a second case pool are provided, the mixing ratio of case information obtained from the first case pool and case information obtained from the second case pool may be determined. If a third case pool or other additional case pools are provided, the mixing ratio may be determined by further including case information obtained from these additional case pools.

[0147] This configuration makes it easier to prioritize the use of highly relevant and operationally useful case study information for new target input information. Furthermore, it allows for adjustment of the balance between store-specific case studies and industry-wide common case studies.

[0148] (Modification 9: Transformation of generation process and generation result) The generation process is not limited to sending a generation request to the external generation processing device 40. For example, the response draft may be generated by a generation model in the user terminal 20, a generation model in the management server 30, a rule-based draft generation process, a template process, a search extension generation process, or a combination thereof. Furthermore, the generation process is not limited to generating a single response draft, but may output multiple candidate drafts, a summary, a confirmation draft, an explanatory text, an internal memo, data for subsequent processing, or other generation results.

[0149] This configuration allows for flexible selection of the response draft generation method and the format of the generated result, depending on the available generation environment, cost, response speed, security requirements, and user verification method.

[0150] (Variation 10: Variation of processing mode and user involvement) The processing modes are not limited to fully automatic mode, automatic mode with confirmation, and proposal mode. For example, the processing modes may include draft save mode, administrator approval mode, legal confirmation mode, assignee mode, regeneration request mode, learning-only mode, and no learning mode. Furthermore, the processing mode may be determined based on the content of the target input information, evaluation range, legal risk score, policy violation score, progress, user privileges, contract plan, platform specifications, and other conditions. As an example, the information processing system 1000 may select fully automatic mode when the legal risk score and policy violation score are low and the user privileges meet predetermined conditions, select automatic mode with confirmation when the risk is moderate, and select proposal mode, administrator approval mode, or legal confirmation mode when the risk is high.

[0151] This configuration allows for adjustment of the autonomy and security of response operations, such as streamlining response processing when the risk is low and requiring verification by users, administrators, or legal personnel when the risk is high. Furthermore, by dynamically switching processing modes based on evaluation tiers, risk scores, and progress, it becomes easier to expand to autonomous agent-based response processing.

[0152] (Variation 11: Variation of the timing of the registration process) The timing for registering confirmed response content as case information is not limited to immediately after the user's confirmation operation. For example, the registration process may be performed after the response content has been sent, after it has been saved as a draft, after administrator approval, after a predetermined period has elapsed, after customer feedback has been obtained after the response, or after a predetermined inspection process has been completed. Alternatively, the registration process may involve provisional registration at the time of confirmation, and then updating to full registration based on the completion of transmission, approval, or inspection.

[0153] This configuration allows for the control of registering the confirmed response as case information, taking into account whether it was actually used, whether it passed administrator verification, or whether it was determined to be free of problems in subsequent inspections.

[0154] (Modification 12: Modification of history management and audit information) The information processing system 1000 may store a history of the determination, registration, use, and generation processes of learning policy information. The history may include information such as whether or not the confirmed response content can be registered, the case pool to which it will be registered, the use weight, the case information used, the generated response text, the content modified by the user, the content of changes to user settings, the processing mode, the date and time of processing, the processor, the target store, and other information.

[0155] This configuration allows for later verification of the conditions under which case information was registered and the weights with which it was used in future response draft generation, thereby enhancing the transparency and auditability of learning policy control.

[0156] Furthermore, the configurations described in the above modified examples do not limit the technical scope of the invention as described in the claims, and can be appropriately selected or combined depending on the embodiment.

[0157] <Note> This embodiment includes the following disclosures.

[0158] (Note 1) An information processing system, Having one or more control units, The control unit, Obtain a response draft generated based on the target input information. Regarding the aforementioned response draft, the confirmed response content is obtained after the user confirms it. The learning control conditions related to the aforementioned target input information, the confirmed response content, or registered case information are acquired. Based on the learning control conditions, the learning policy information is determined, including whether the confirmed response content can be registered as case information, the registration destination of the case information, and the usage weight for future use of the case information. Based on the learning policy information, the process controls the registration of the confirmed response content as case information and the generation of a new response draft using the registered case information. Information processing system.

[0159] According to this note, compared to uniformly registering and using confirmed response content as case information, it is possible to control whether or not to register it as case information, where to register it, and the weight of its use, according to the learning control conditions.

[0160] (Note 2) The information processing system described in Appendix 1, The learning policy information includes registration eligibility conditions that determine whether the confirmed response content can be registered as case information, pool designation conditions that determine the destination for registration of the case information, and usage weight conditions that determine the contribution of the case information when it is used in the future. The control unit, Based on the registration eligibility conditions and the pool designation conditions, the registration process is controlled to register the confirmed response content as the case information. Based on the aforementioned usage weighting conditions, the usage process for using the registered case information in response text generation is controlled. When generating a response text corresponding to the new target input information, the generation process is controlled using the case information selected and weighted based on the learning policy information. Information processing system.

[0161] According to this addendum, the registration process for confirmed response content, the process of using registered case information, and the generation process for new response drafts can be controlled more specifically using registration eligibility conditions, pool designation conditions, and usage weight conditions.

[0162] (Note 3) The information processing system described in Appendix 2, The learning control conditions include an evaluation score associated with the target input information, The control unit determines the registration eligibility conditions, pool designation conditions, and usage weight conditions based on the evaluation tier to which the evaluation score belongs. Information processing system.

[0163] According to this note, the registration of confirmed response content as case information, the registration destination, and the degree of contribution when used in the future can be adjusted according to the evaluation range to which the evaluation score belongs.

[0164] (Note 4) An information processing system as described in Appendix 2 or Appendix 3, The learning control conditions include the elapsed time from the registration date or confirmation date of the registered case information, The control unit determines the usage weight condition for reducing the usage weight of the registered case information in accordance with the increase in elapsed time, When acquiring the registered case information, the effective utilization weight of the case information is dynamically calculated based on the utilization weight conditions. Information processing system.

[0165] This addendum reduces the impact of older registered case information with older registration or confirmation dates, making it easier to generate response drafts that align with relatively new operational policies. Furthermore, by dynamically calculating effective utilization weights when case information is used or acquired, weighting can be applied that reflects the latest elapsed time at the time of use, without having to update the case information or utilization weight information stored in the memory each time.

[0166] (Note 5) An information processing system described in any one of the appendices 2 to 4, The learning control conditions include legal risk scores and policy violation scores related to the definitive response content. The control unit determines, based on the legal risk score and the policy violation score, the registration eligibility conditions for registering the confirmed response content as case information. Information processing system.

[0167] This addendum helps to prevent the accumulation of definitive responses that may pose legal risks or violate policies as case information used in generating future response drafts.

[0168] (Note 6) An information processing system described in any one of the appendices 2 to 5, The control unit calculates a learning suitability score based on the evaluation score, elapsed time, legal risk score, and policy violation score included in the learning control conditions. Based on the aforementioned learning fit score, the registration eligibility criteria and the usage weighting criteria are determined. Information processing system.

[0169] According to this note, by comprehensively considering multiple learning control conditions, it is possible to determine whether or not to register definitive response content and to determine its usage weight.

[0170] (Note 7) An information processing system described in any one of the appendices 2 to 6, The control unit calculates the difference between the draft response and the finalized response content. The difference amount is used, together with the learning control conditions, as one of several determination elements for determining the registration eligibility condition and the usage weight condition. Information processing system.

[0171] According to this note, the extent of modifications made by the user to the response draft can be considered in combination with evaluation scores, elapsed time, legal risk scores, policy violation scores, and other learning control conditions, making it easier to appropriately determine the usefulness of the finalized response content as case information.

[0172] (Note 8) An information processing system described in any one of the appendices 2 to 7, The aforementioned case information is managed by dividing it into multiple case pools, which include at least a first case pool associated with stores and a second case pool associated with industries. The control unit determines the pool designation conditions for determining the registration destination of the confirmed response content from among the plurality of case pools based on the learning control conditions. Information processing system.

[0173] According to this addendum, store-specific response trends and industry-wide response trends can be managed separately. Furthermore, if additional case pools are created that correspond to service types or other additional classifications, response trends corresponding to those additional classifications can also be managed separately.

[0174] (Note 9) The information processing system described in Appendix 8, The control unit, When registering the confirmed response content in the second case pool, the first industry classification information transmitted from the user terminal is acquired. When generating a response draft corresponding to the aforementioned new target input information, the second industry classification information is obtained based on the contract information, store information, or account information of the authenticated account. When the first industry classification information and the second industry classification information belong to different classification systems, the system controls registration to the same second case pool and retrieval of case information from the same second case pool by converting the first industry classification information and the second industry classification information into common industry identification information or pool identification information of the second case pool based on industry mapping information. Information processing system.

[0175] According to this note, even if the industry information transmitted from the user's terminal during registration and the industry information derived from the contract information of the authenticated account during acquisition belong to different classification systems, the same second case pool can be matched based on the industry mapping information, making it easier to appropriately register and use case information common to all industries.

[0176] (Note 10) The information processing system described in Appendix 8, When the control unit generates a response statement corresponding to the new target input information, it determines the mixing ratio of the case information obtained from the multiple case pools based on the learning policy information. Information processing system.

[0177] According to this addendum, depending on the new target input information, it will be easier to utilize store-specific examples, industry-wide examples, and examples from additional example pools, if any, in an appropriate ratio.

[0178] (Note 11) An information processing system described in any one of the appendices 1 to 10, The control unit receives user settings regarding whether the confirmed response content can be registered, the registration destination, and the usage weight. The learning policy information is updated based on the user settings. Information processing system.

[0179] According to this note, the content of the registration and use control of case information can be adjusted according to the operational policies of each store, business, and account.

[0180] (Note 12) An information processing system described in any one of the appendices 1 through 11, The control unit determines a processing mode from among a fully automatic mode that executes from the presentation of a response draft to transmission, an automatic mode with confirmation that executes transmission after user approval, and a proposal mode that only presents a response draft, based on the learning control conditions. Information processing system.

[0181] According to this note, the degree of user involvement in post-response text generation processing can be adjusted according to the learning control conditions.

[0182] (Note 13) An information processing method performed by an information processing system, Obtain a response draft generated based on the target input information. Regarding the aforementioned response draft, the confirmed response content is obtained after the user confirms it. The learning control conditions related to the aforementioned target input information, the confirmed response content, or registered case information are acquired. Based on the learning control conditions, the learning policy information is determined, including whether the confirmed response content can be registered as case information, the registration destination of the case information, and the usage weight for future use of the case information. Based on the learning policy information, the process controls the registration of the confirmed response content as case information and the generation of a new response draft using the registered case information. Information processing methods.

[0183] According to this note, it is not limited to the device configuration, but it is possible to understand the processing procedure itself that controls whether or not the confirmed response content can be registered, the registration destination, and the usage weight.

[0184] (Note 14) It is a program, Computers, A program to function as an information processing system as described in any one of the appendices 1 through 12.

[0185] According to this appendix, each of the configurations described in appendices 1 to 12 can be implemented on a computer, and therefore can serve as the basis for obtaining rights corresponding to program products or software distribution formats. [Explanation of Symbols]

[0186] 20 user terminals 30 Management Servers 40 External generation processing device 50 Platform Servers 60 Networks 210 Control Unit 220 Storage section 230 Input section 240 Output section 250 Communications Department 260 Internal Bus 1000 Information Processing Systems

Claims

1. An information processing system, Having one or more control units, The control unit, Obtain a response draft generated based on the target input information. Regarding the aforementioned response draft, the finalized response content is obtained after a confirmation or editing operation by the user. Based on the aforementioned target input information, the confirmed response content, or the information contained in the registered case information, or the information associated therewith, learning control conditions related to the aforementioned target input information, the confirmed response content, or the registered case information are acquired. The learning control condition includes at least one of the following: evaluation information associated with the target input information, the content of the target input information, the content of the confirmed response, the confirmation date of the confirmed response, store information, business information, account information, industry information, platform type information, or condition information used for generating response text or managing case information. Based on the learning control conditions, the learning policy information is determined, including whether or not the confirmed response content can be registered as example information, the registration destination of the example information if the confirmed response content is registered as example information, and the usage weight for future use of the registered example information. Based on the learning policy information, the process controls the registration of the confirmed response content as case information and the generation of a new response draft using the registered case information. Information processing system.

2. The information processing system according to claim 1, The learning policy information includes registration eligibility conditions that determine whether the confirmed response content can be registered as case information, pool designation conditions that determine the destination for registration of the case information, and usage weight conditions that determine the contribution of the case information when it is used in the future. The control unit, Based on the registration eligibility conditions and the pool designation conditions, the registration process is controlled to register the confirmed response content as the case information. Based on the aforementioned usage weighting conditions, the usage process for using the registered case information in response text generation is controlled. When generating a response text corresponding to the new target input information, the generation process is controlled using the case information selected and weighted based on the learning policy information. Information processing system.

3. The information processing system according to claim 2, The learning control conditions include an evaluation score associated with the target input information, The control unit determines the registration eligibility conditions, pool designation conditions, and usage weight conditions based on the evaluation tier to which the evaluation score belongs. Information processing system.

4. The information processing system according to claim 2, The learning control conditions include the elapsed time from the registration date or confirmation date of the registered case information, The control unit determines the usage weight condition for reducing the usage weight of the registered case information in accordance with the increase in elapsed time, When acquiring the registered case information, the effective utilization weight of the case information is dynamically calculated based on the utilization weight conditions. Information processing system.

5. The information processing system according to claim 2, The learning control conditions include legal risk scores and policy violation scores related to the definitive response content. The control unit determines, based on the legal risk score and the policy violation score, the registration eligibility conditions for registering the confirmed response content as case information. Information processing system.

6. The information processing system according to claim 2, The control unit calculates a learning suitability score based on the evaluation score, elapsed time, legal risk score, and policy violation score included in the learning control conditions. Based on the aforementioned learning fit score, the registration eligibility criteria and the usage weighting criteria are determined. Information processing system.

7. The information processing system according to claim 2, The control unit calculates the difference between the draft response and the finalized response content. The difference amount is used, together with the learning control conditions, as one of several determination elements for determining the registration eligibility condition and the usage weight condition. Information processing system.

8. The information processing system according to claim 2, The aforementioned case information is managed by dividing it into multiple case pools, which include at least a first case pool associated with stores and a second case pool associated with industries. The control unit determines the pool designation conditions for determining the registration destination of the confirmed response content from among the plurality of case pools based on the learning control conditions. Information processing system.

9. The information processing system according to claim 8, The control unit, When registering the confirmed response content in the second case pool, the first industry classification information transmitted from the user terminal is acquired. When generating a response draft corresponding to the aforementioned new target input information, the second industry classification information is obtained based on the contract information, store information, or account information of the authenticated account. When the first industry classification information and the second industry classification information belong to different classification systems, the system controls registration to the same second case pool and acquisition of case information from the same second case pool by converting the first industry classification information and the second industry classification information into common industry identification information or pool identification information of the second case pool based on industry mapping information. Information processing system.

10. The information processing system according to claim 8, When the control unit generates a response statement corresponding to the new target input information, it determines the mixing ratio of the case information obtained from the multiple case pools based on the learning policy information. Information processing system.

11. The information processing system according to claim 1, The control unit receives user settings regarding whether the confirmed response content can be registered, the registration destination, and the usage weight. The learning policy information is updated based on the user settings. Information processing system.

12. The information processing system according to claim 1, The control unit determines a processing mode from among a fully automatic mode that executes from the presentation of a response draft to transmission, an automatic mode with confirmation that executes transmission after user approval, and a proposal mode that only presents a response draft, based on the learning control conditions. Information processing system.

13. An information processing method performed by an information processing system, Obtain a response draft generated based on the target input information. Regarding the aforementioned response draft, the finalized response content is obtained after a confirmation or editing operation by the user. Based on the aforementioned target input information, the confirmed response content, or the information contained in the registered case information, or the information associated therewith, learning control conditions related to the aforementioned target input information, the confirmed response content, or the registered case information are acquired. The learning control condition includes at least one of the following: evaluation information associated with the target input information, the content of the target input information, the content of the confirmed response, the confirmation date of the confirmed response, store information, business information, account information, industry information, platform type information, or condition information used for generating response text or managing case information. Based on the learning control conditions, the learning policy information is determined, including whether or not the confirmed response content can be registered as example information, the registration destination of the example information if the confirmed response content is registered as example information, and the usage weight for future use of the registered example information. Based on the learning policy information, the process controls the registration of the confirmed response content as case information and the generation of a new response draft using the registered case information. Information processing methods.

14. Computers, A program for functioning as an information processing system according to any one of claims 1 to 12.

Images