Information processing system, method for processing information, and information processing program

Abstract

To provide an information processing system, a method for processing information, and an information processing program for automatically generating candidates of replies to electronic mails and for achieving improvement of work efficiency of a user.SOLUTION: The information processing system performs the steps of receiving, for a first electronic mail, a reply candidate generation command transmitted from a personal terminal; a function for obtaining the first electronic mail and a second electronic mail related to the first electronic mail; generating a reply candidate for the first electronic mail on the basis of the first electronic mail and the second electronic mail; and transmitting the reply candidate to the personal terminal.SELECTED DRAWING: Figure 3

Description

[Technical Field]

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

[0002] With the advancement of information technology, e-mail has become an essential means of communication. As a result, people spend a lot of time using e-mail in their work, and in order to improve work efficiency, technology has been developed to support the generation of reply messages when receiving e-mails.

[0003] For example, Patent Document 1 discloses a technology that uses a machine learning learning model to assist in generating a draft of part of the body of a reply email to an incoming email when part of the body of the incoming email is input. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2024-33038 Summary of the Invention [Problem to be solved by the invention]

[0005] According to the technology described in Patent Document 1, a user can obtain at least a partial draft of the body of a reply email for a received email. However, while the technology described in Patent Document 1 uses machine learning technology, the learning model uses the relationship between the sender and recipient of the email as basic information, and does not generate a reply proposal that takes into account the specific content of the email. As a result, the reply proposal may deviate from the reply email that the user envisions, or only a partial draft may be output, which ultimately requires the user to revise the reply proposal, thereby preventing maximum business efficiency.

[0006] In order to solve the above-mentioned problems, the object of the present invention is to provide an information processing device, an information processing system, an information processing method, and an information processing program that enable business efficiency to be improved by efficiently and effectively generating reply proposals for received emails. [Means for solving the problem]

[0007] An example of an information processing device according to the present invention is receiving a reply draft generation command sent from a personal device in relation to the first email; acquiring the first email and a second email related to the first email; generating a reply proposal to the first email based on the first email and the second email; The proposed reply is sent to the personal terminal.

[0008] An example of an information processing method according to the present invention includes: receiving, by the computer, a reply generation command sent from the personal terminal in relation to the first email; A computer acquires the first email and a second email related to the first email; generating a reply proposal to the first email based on the first email and the second email by a computer; a computer sending the proposed reply to the personal terminal; Equipped with.

[0009] An example of an information processing program according to the present invention causes a computer to function as the above-described information processing system. [Effects of the Invention]

[0010] According to the present invention, the technology relating to the information processing system, information processing method, and information processing program described in the following embodiments makes it possible to efficiently and effectively generate reply proposals for received emails, thereby achieving improved business efficiency. [Brief explanation of the drawings]

[0011] [Figure 1] A diagram showing the configuration of an information processing system [Figure 2] FIG. 1 shows a configuration of an information processing device. [Figure 3] Flowchart showing a process for generating reply suggestions by an information processing system [Figure 4] The screen displayed when the mailer executes the reply suggestion function [Figure 5] 1 is a flowchart showing the process performed by an information processing system from receiving a reply proposal generation command to displaying a reply proposal. [Figure 6] 1 is a flowchart showing the process from when an information processing system receives a question generation command to when it displays a reply proposal. [Figure 7] In the reply proposal generation process after the summary generation process, the prompt sent from the reply proposal generation support server to the generation AI server [Figure 8] In the reply proposal generation process, prompts sent from the reply proposal generation support server to the generation AI server [Figure 9] (a) is the screen that displays the generated reply plan, (b) is the screen that displays the questions and answer candidates when regenerating a reply plan, and (c) is the screen that is displayed when generating additional questions when regenerating a reply plan. [Figure 10] 1 is a flowchart showing the processing performed by an information processing system from receiving a command to generate a summary to displaying the summary. [Figure 11] In the summary generation process, prompts are sent from the reply generation support server to the generation AI server. [Figure 12] A screen showing the generated summary DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, examples of embodiments of the present invention will be described with reference to the drawings.

[0013] FIG. 1 is a diagram illustrating a configuration of an information processing system 1 according to an embodiment of the present invention. The information processing system 1 is a system configured by connecting a terminal 11 (personal terminal) that is a personal terminal operated by a user and sends and receives emails (hereinafter, sometimes simply referred to as "email"), a mail server 12 that delivers emails sent and received by the terminal 11, a reply proposal generation support server 13 that supports the generation of reply emails to emails received by the terminal 11, and a generation AI server 14 (AI: Artificial Intelligence) that aims to generate new content such as text using technologies such as deep learning and large datasets, all of which are communicatively connected via a network NW. The network NW is configured by the Internet via wireless communication such as Wi-Fi (registered trademark) and wired communication. The information processing system 1 executes the information processing method described herein.

[0014] The terminal 11 is equipped with a mailer, which is an application program (hereinafter, an application program may be simply referred to as an "application") for creating, viewing, sending, receiving, and managing emails, and the user sends and receives emails via the mailer. The mailer may take the form of, for example, email software installed on the terminal 11 as an application and accessing a mail server. Alternatively, the mailer may take the form of cloud email, which runs on a browser for browsing the Internet and accesses the mail server via a web server. While the present embodiment will be described using the cloud email format as an example, the present invention can also be realized with email software in the same way as with cloud email. A reply assistance application that works in conjunction with the reply suggestion generation assistance server 13 is also installed on the terminal 11. The reply assistance application operates, for example, as an add-in for the mailer or browser. The reply assistance application may be integrated into the mailer, i.e., the mailer itself may have a function for working in conjunction with the reply suggestion generation assistance server 13.

[0015] In one embodiment, the reply suggestion generation support server 13 may control the reply support application. For example, the reply suggestion generation support server 13 may send instructions to the terminal 11 regarding displaying a GUI related to the reply support application (such as the reply support GUI 41 shown in FIG. 4 ), accepting input from the user to the GUI, etc., and the terminal 11 may receive these instructions and execute them to realize the functions of the reply support application.

[0016] In this embodiment, received mail refers to an email that can be checked via the mailer of terminal 11 and has been sent to the user, and reply mail refers to an email that is a reply to the received mail and is sent by the user via the mailer of terminal 11. In this embodiment, a case where emails are sent and received one-to-one will be described as a specific example, but the present invention can also be realized in the case of one-to-many and many-to-many cases.

[0017] 2 is a diagram showing the configuration of an information processing device according to an embodiment of the present invention. Although the terminal 11, mail server 12, reply suggestion generation support server 13, and generation AI server 14 have different roles in the information processing system 1 of this embodiment, they are all information processing devices with the same hardware configuration.

[0018] The information processing device comprises a control unit 21 consisting of a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), etc.; a storage device 22 consisting of a volatile memory and a non-volatile memory, etc.; a communication device 23 having a communication processing circuit and communicating data with the outside; an input device 24 consisting of a keyboard, mouse, camera, microphone, etc.; and an output device 25 consisting of a display (including one with a touch panel function) and speakers, etc., and transmits and receives data via a system bus indicated by an arrow.

[0019] The control unit 21 controls the information processing device in accordance with a predetermined program. This program causes the computer to function as the information processing system described in this specification. The program is stored in the storage device 22. The storage device 22 includes a program area into which executable files are deployed when the program is executed. The information processing device can also have its functions expanded by downloading new application programs from an application server (not shown) via a network. In this case, the new application programs are also stored in the storage device 22. The communication device 23 is composed of a LAN (Local Area Network) communication interface, a mobile telephone communication network communication unit, a close-proximity wireless communication unit, etc.

[0020] The terminal 11 may be any information processing device capable of bidirectionally transmitting and receiving data to and from other information processing devices, such as a PC (Personal Computer), a mobile phone, a smartphone, a tablet terminal, a PDA (Personal Digital Assistant), a digital camera, a portable game console, or other portable digital devices. In this embodiment, a PC will be used as a specific example.

[0021] The mail server 12, reply proposal generation support server 13, and generation AI server 14 are information processing devices that are always connected to the network.

[0022] The generation AI server 14 stores, for example, an LLM (large-scale language model), and can input natural language data into the LLM to obtain its output. The LLM is a trained model that has been trained using, for example, a large amount of text data to output appropriate natural language responses to natural language inputs, and various known models can be used. Known models include, but are not limited to, GPT, BERT, PaLM, and Llama. The generation AI server 14 can input input data (including prompts and reference data) received from an external source into the LLM and transmit the corresponding output from the LLM to the outside. In this embodiment, the input and output to the LLM is Japanese, but other languages ​​may be used instead of or in addition to Japanese.

[0023] FIG. 3 is a flowchart showing a process for generating reply suggestions by the information processing system 1 according to the embodiment of the present invention.

[0024] First, the user starts the execution of the mailer on the terminal 11. In response, the reply assistance application starts to run and its screen is displayed. After that, the user selects a specific email (first email) and starts composing a reply email to it.

[0025] The process branches depending on whether the first email forms a thread with another email (second email) (step S1). A thread here refers to a group of emails that are related to the content of the received email.

[0026] As a specific example, whether a given email and another email form a thread is determined by referring to one or more of the email body, the email header information (e.g., Subject, Message-ID, Reply-To, In-Reply-To, References), thread identification information specific to the mailer, etc. When the subject is used for the determination, for example, if the subject of a first email begins with the character string "Re:", emails with the same subject except for the character string Re: are included in the thread. In this way, the second email can be configured to be identified based on the header information of the first email. The thread determination may be performed by any of the terminal 11, the mail server 12, or the reply suggestion generation support server 13.

[0027] If a thread exists, proceed to step S2, and if no thread exists, proceed to step S8. Note that, for the sake of simplicity, this specification may refer to the first email forming a thread together with other emails as "having a thread," and the first email forming an independent thread on its own as "not having a thread."

[0028] Next, terminal 11 enters a state of waiting for an input from the user as to whether to generate a reply proposal or a summary for the first email (step S2). If an input to generate a reply proposal is received from the user, the process proceeds to step S3. If an input to generate a summary is received from the user, the process proceeds to step S5. Although not shown, an input to terminate execution of the reply assistance application may be received in this state, or execution of the reply assistance application may be terminated automatically if no input is received for a certain period of time. The same applies below as long as terminal 11 is in a state of receiving input from the user.

[0029] Next, when terminal 11 receives an input to generate a reply suggestion in step S2, it sends a command to generate a reply suggestion for the first email (a reply suggestion generation command) to reply suggestion generation support server 13. Having received the reply suggestion generation command, reply suggestion generation support server 13 generates a reply suggestion in cooperation with mail server 12 and generation AI server 14 (step S3). This process will be described in detail using FIG. 4 etc. Having generated a reply suggestion, reply suggestion generation support server 13 sends it to terminal 11. Having received the reply suggestion, terminal 11 displays the reply suggestion on the screen of its mailer (step S4). Note that the "mailer screen" refers to the screen that is displayed on the display device of terminal 11 when terminal 11 executes the mailer.

[0030] On the other hand, if terminal 11 accepts an input to generate a summary in step S2, terminal 11 sends a command (summary generation command) to reply suggestion generation support server 13 to generate summaries for other emails in the same thread as the first email. Having accepted the command to generate a summary, reply suggestion generation support server 13 cooperates with mail server 12 and generation AI server 14 to generate a summary (step S5). This process will be described in detail using FIG. 10 etc. Having generated the summary, reply suggestion generation support server 13 sends the summary to terminal 11. Having received the summary, terminal 11 displays the summary on the screen of its mailer (step S6).

[0031] Next, terminal 11 waits for the user to input whether to generate a reply plan for the first email or to input additional information for generating a reply plan (step S7). If the user input to generate a reply plan is received, terminal 11 transmits a reply plan generation command for the first email to reply plan generation assistance server 13, and proceeds to the processing of step S3 described above.

[0032] In this case, however, since a summary for the first e-mail has already been generated, the command can be to generate a reply plan that takes the summary into account, and the process of generating the summary can be omitted. For example, the reply plan generation support server 13 receives a reply plan generation command for the first e-mail sent from the terminal 11, and in response to this reply plan generation command, generates a reply plan for the first e-mail based on the first e-mail and the summary that has already been generated.

[0033] In step S7, if an input to generate a question is received from the user, a command to generate a question and answer candidates for inputting additional information for the reply suggestion is sent to the reply suggestion generation support server 13, and the process proceeds to step S10.

[0034] After terminal 11 displays a suggested reply to the first email in step S4, terminal 11 waits for the user to input whether to generate a question and candidate answers to regenerate the suggested reply, or to transcribe the displayed suggested reply as the text of a reply email (step S9).

[0035] When the terminal 11 receives an input from the user to generate a question and answer candidates for regenerating a reply suggestion, it sends a command to generate a question and answer candidates to the reply suggestion generation support server 13. Upon receiving this command, the reply suggestion generation support server 13 works with the mail server 12 and the generation AI server 14 to generate a question and answer candidates (step S10). The question and answer candidates are a question to the user to input additional information, and answer candidates for the question. The reply suggestion generation support server 13 sends the question and answer candidates to the terminal 11.

[0036] Terminal 11 receives the question and answer candidates and displays them on the screen of the mailer (step S11). Terminal 11 then receives answers to the question from the user (step S12).

[0037] On the other hand, if step S9 accepts an input from the user to transcribe the reply proposal as the reply text, terminal 11 generates the reply text for the received email by transcribing the reply proposal and displays it on the screen of the mailer (step S16). Here, if the reply text has already been entered before the transcription process, the reply text may be added after the text already entered, or the text already entered may be deleted and only the reply proposal may be displayed as the reply text.

[0038] After receiving an answer to the question from the user in step S12, terminal 11 waits for the user to input whether to generate a reply proposal by referring to the input answer, or to input further information and regenerate the question and answer candidates (step S13).

[0039] When terminal 11 receives input from the user to generate a reply suggestion, it transmits a command to generate a reply suggestion to reply suggestion generation support server 13. Having received the command to generate a reply suggestion, reply suggestion generation support server 13 generates a reply suggestion in cooperation with mail server 12 and generation AI server 14 (step S14). Having generated a reply suggestion, reply suggestion generation support server 13 transmits the reply suggestion to terminal 11. Having received the reply suggestion, terminal 11 displays the reply suggestion on the screen of its mailer (step S15). Terminal 11 then proceeds to step S9, where it receives input again.

[0040] On the other hand, if an input to regenerate a question and answer candidates by inputting additional information from the user is received in step S13, the additional question and answer are generated by performing the same process as when an input to input additional information is received in step S9. The series of processes from steps S10 to S15 will be described in detail using Figure 6, etc. In this way, by inputting additional information to the displayed reply plan and generating a new reply plan taking that information into account, it is possible to generate a more optimal reply plan.

[0041] If there is no thread for the first email in step S1, the terminal 11 enters a state of waiting for an input from the user to generate a reply proposal for the first email (step S8). If an input to generate a reply proposal is received from the user, the terminal 11 transmits a command to generate a reply proposal for the first email to the reply proposal generation assistance server 13, and proceeds to the processing of step S3 described above. On the other hand, if there is no input, no particular processing is executed (the reply assistance application may be terminated). The terminal 11 may remain in an input standby state until an input is received.

[0042] 4 shows a screen displayed when a mailer according to an embodiment of the present invention executes the reply suggestion generation function. Specifically, this is a screen displayed on a display device (e.g., a monitor) of terminal 11 when terminal 11 is ready to accept input from the user in step S2.

[0043] On the display, a GUI (reply support GUI 41) of a reply support application that works in conjunction with reply plan generation support server 13 is superimposed on the screen that is displayed when the mailer is launched. For example, when a specific email (first email) is selected by the user on terminal 11 and a screen for creating a reply to that email is displayed, reply support GUI 41 may switch to displaying the content shown in Fig. 4.

[0044] In the reply assistance application, in step S2, the terminal 11 accepts input from the user to generate a suggested reply or a summary. For this purpose, buttons for this purpose are displayed in the reply assistance GUI 41. In the example of FIG. 4, the button labeled "Generate" in the reply assistance GUI 41 is the reply suggestion generation button, and the button labeled "Summary" is the summary generation button. A reply suggestion generation command is input by operating the reply suggestion generation button, and a summary generation command is input by operating the summary generation button. Note that if the email for which a reply suggestion is to be generated does not have a thread, the button for accepting input to generate a summary does not need to be displayed.

[0045] 4, when a specific email (first email) is selected in the mailer, or when the contents of a specific email are displayed, the reply assistance GUI 41 includes only a reply proposal generation button and a summary generation button as input buttons. This configuration allows the user to instruct the generation of a reply proposal or a summary with a single operation (for example, just one click) without having to check the details of the GUI or perform complex operations to enter information.

[0046] 4, the reply assistance GUI 41 includes a reply suggestion generation button and a summary generation button as input buttons, but the configuration of the input buttons can be modified in various ways. For example, it may include only the reply suggestion generation button but not the summary generation button, or it may include only the summary generation button but not the reply suggestion generation button. Furthermore, other buttons may be included instead of or in addition to these buttons.

[0047] Figure 5 is a flowchart showing the processing performed by an information processing system 1 according to an embodiment of the present invention from receiving a reply proposal generation command to displaying a reply proposal, and explains in detail the series of processing steps S2 to S4 in the flowchart shown in Figure 3.

[0048] In step S2, the terminal 11 accepts an input from the user to generate a reply suggestion. Next, the terminal 11 transmits a reply suggestion generation command to the reply suggestion generation support server 13 (step S31). As described above, this reply suggestion generation command is transmitted from the terminal 11 in relation to the first email. In response to receiving the reply suggestion generation command, the reply suggestion generation support server 13 transmits a command to the mail server 12 to acquire the first email (the first email that is the subject of the reply suggestion generation command) and all other emails included in the thread of the first email (second emails, i.e., other emails related to the first email) (step S32). Upon receiving this command, the mail server 12 transmits the first email and the second email to the reply suggestion generation support server 13, which then receives and acquires them (step S33).

[0049] However, if the input for generating a reply proposal has been received via step S8 (i.e., if the second email does not exist), only the first email is acquired. Also, if the input for generating a reply proposal has been received via step S7 (i.e., if a summary of the second email has already been generated), only the first email may be acquired. Note that if a summary of the second email has not yet been generated at this point, the summary generation process described with reference to FIG. 10 etc. may be performed at this point.

[0050] Next, the reply suggestion generation support server 13 sends an email reply suggestion generation command to the generation AI server 14 (step S34). The reply suggestion generation command includes information indicating to the generation AI server 14 that the content to be generated is the generation of a reply message for an email, and information related to the email acquired by the reply suggestion generation support server 13 in step S33. For example, the command may include information representing the contents of the first email and the second email. Or, for example, the command may include information representing the contents of the first email and a summary of the second email. Or, for example, the command may include information representing the summary of the first email and a summary of the second email.

[0051] Upon receiving the reply plan generation command, the generation AI server 14 generates a reply plan based on the received information and transmits the generated reply plan to the reply plan generation support server 13 (step S35).

[0052] In this way, the reply plan generation support server 13 generates a proposed reply to the first email using the generation AI server 14 based on the first email and the second email. More specifically, when generating a proposed reply, the reply plan generation support server 13 sends a command to the generation AI server 14 to generate a proposed reply for the first email and the second email, receives the proposed reply generated by the generation AI server 14 from the generation AI server 14, and generates a proposed reply based on the reply. The reply plan generation support server 13 receives the proposed reply and sends it to the terminal 11 (step S36). The terminal 11 displays the received proposed reply on the screen of its mailer (step S4).

[0053] Figure 6 is a flowchart showing the processing from receiving a question generation command by an information processing system 1 according to an embodiment of the present invention to displaying reply suggestions, and explains the details of the series of processing steps S9 to S15 in the flowchart shown in Figure 3.

[0054] In step S9, the terminal 11 receives from the user an instruction to generate a question and answer candidates for regenerating a reply plan. Next, the terminal 11 transmits the question generation instruction to the reply plan generation support server 13 (step S101). The reply plan generation support server 13, having received the question generation instruction, transmits to the generation AI server 14 an instruction to generate a question and answer candidates for the user to enter additional information required to regenerate a reply plan (step S102). Details of the instruction transmitted in step S102 will be explained using FIG. 8 and other figures. Next, the generation AI server 14 generates a question and answer candidates and transmits them to the reply plan generation support server 13 (step S103). The reply plan generation support server 13, having received the question and answer candidates, transmits them to the terminal 11 (step S104). In this way, the reply plan generation support server 13 receives the question generation instruction transmitted from the terminal 11, generates a question for the reply plan and answer candidates for the question in accordance with the question generation instruction, and transmits the question and answer candidates to the terminal 11.

[0055] Terminal 11 displays the accepted question and answer candidates on the screen of the mailer (step S11). Terminal 11 then accepts an answer from the user (step S12). Next, terminal 11 accepts a command to generate a reply proposal from the user (step S13). Note that this assumes that the process branches from step S13 to S14 in the flowchart of FIG. 3.

[0056] When the terminal 11 receives the command to generate a reply suggestion, it transmits the command to the reply suggestion generation support server 13 (step S141). Having received the command to generate a reply suggestion, the reply suggestion generation support server 13 transmits the command to the generation AI server 14 to generate a reply suggestion (step S142). The reply suggestion generation command includes information indicating to the generation AI server that the content to be generated is the generation of a reply message for an email, the content of the email acquired by the reply suggestion generation support server 13, and the answer and question related to the answer received from the user in step S12. In this way, the reply suggestion generation support server 13 transmits commands to the generation AI server 14 to generate reply suggestions for the first email and the second email.

[0057] Having received the command to generate a suggested reply, the generation AI server 14 generates a suggested reply based on the received information and transmits the generated suggested reply to the reply proposal generation support server 13 (step S143). In this way, the reply proposal generation support server 13 receives the suggested reply generated by the generation AI server 14 from the generation AI server 14. In this way, the reply proposal generation support server 13 can be said to generate a suggested reply using the generation AI server 14. The reply proposal generation support server 13 receives and generates a suggested reply, and transmits it to the terminal 11 (step S144). The terminal 11 displays the received suggested reply on the screen of its mailer (step S15).

[0058] An example of a prompt for the generation AI server 14 is described below.

[0059] 7 shows an example of a prompt sent from the reply plan generation support server 13 to the generation AI server 14 in the reply plan generation process according to an embodiment of the present invention. More specifically, this is a statement sent from the reply plan generation support server 13 to the generation AI server 14 in step S34 to instruct the generation AI server 14 to generate a reply plan.

[0060] The prompt in Figure 7 shows how to generate suggested replies for a thread containing three emails, such as the following:

[0061] The email sent on the earliest date and time (the second email) was sent from Mr. XX of Co., Ltd. A to Mr. ○○ of Co., Ltd. B, and the text of the email is as follows: "B ○○ Co., Ltd. My name is XX from A Co., Ltd. Our company is currently running an on-premise file management web application and is considering migrating it to the AWS cloud environment. Considering your company's extensive experience and advanced technical capabilities, I am writing to you in the hope that you will support us in carrying out this project. The current application mainly has the following features: Language: Python Database: PostgreSQL Number of users: Approximately 1,000 Data volume: Approximately 5TB During the transition, we want to pay particular attention to the following points: ·Safe data transfer Minimizing system downtime Optimization of operational costs after migration Could you please suggest a series of processes from your company, from the initial investigation to the implementation of the migration and operational support? Also, could you please give us a rough estimate and a rough idea of ​​the schedule? Furthermore, we would like to schedule an initial meeting to discuss the details of this project. Could you please suggest some dates and times that would be convenient for your company? We will do our best to accommodate as many requests as possible. We apologize for bothering you during your busy schedule, but we would appreciate your consideration. A Co., Ltd. Information Systems Department ××」

[0062] The next email sent (the second email) was sent from Mr. XX of Co., Ltd. B to Mr. XX of Co., Ltd. A, and the text of the email is as follows: "A XX Co., Ltd. This is ○○ from B Co., Ltd. Thank you very much for contacting us. Thank you very much for your inquiry regarding the migration of your company's file management web application to AWS and for your proposal for an initial meeting. We are very honored to be involved in your important project. We would like to conduct an initial review based on the information you have provided. We would like to schedule an initial meeting to exchange more detailed information about your current system configuration and migration requests, and to provide you with a concrete proposal. We have listed the dates and times we can accommodate below. If any of these dates and times are convenient for you, please let us know. January 20th (Thu) 13:00~15:00 January 22nd (Fri) 10:00~12:00 January 25th (Tue) 15:00~17:00 We are also assuming that the meeting will be held online, but we would be happy to consider holding a face-to-face meeting if desired. We would also appreciate your feedback on the format of the meeting. We will continue to work with you to realize optimal cloud solutions. We apologize for the inconvenience, but we appreciate your consideration. B Co., Ltd. Cloud Solutions Department ○○"

[0063] The last email (first email) was sent from Mr. XX of Co., Ltd. A to Mr. ○○ of Co., Ltd. B, and the text of the email is as follows: "B ○○ Co., Ltd. Thank you for your reply. This is XX from A Co., Ltd. Of the dates you have proposed, Saturday, January 22nd, 10:00-12:00 would be convenient for us. We would prefer to hold the meeting online. If you have any recommended tools, such as Zoom, Teams, or Google Meet, please let us know. At the meeting, we would like to discuss the details of the current system, specific concerns regarding the migration, and a rough outlook on the budget and schedule. If you need any materials in advance, please let us know. We look forward to working with you to ensure a smooth and efficient cloud migration, and look forward to hearing from you to set up a meeting. A Co., Ltd. Information Systems Department ××」

[0064] 7 makes it clear to the generation AI server 14 that its role is to generate a reply proposal for an email. This part corresponds to the first four lines in FIG.

[0065] In FIG. 7, the "thread synopsis" includes a summary of the second email, and if there are multiple second emails, it includes each summary individually (or a comprehensive summary). In FIG. 7, the "last email" includes the content of the first email (including at least the body, and may also include other information such as the subject). "Constraints" refer to conditions that must be observed as constraints when generating a reply proposal, and include, for example, conditions for ensuring a suitable format for the reply proposal. Specific constraints include that no explanation of the wording in the body of the email is required, that a signature is not required, etc. Fixed text can be stored in advance as "constraints."

[0066] The method for generating a summary based on the second email is not particularly limited, but it can be generated, for example, by the process described in relation to Figure 10. Note that although this embodiment uses the generation AI server 14, other embodiments are also possible in which the summary is generated via a trained model, deterministic rules, etc. other than the generation AI.

[0067] 8 shows a prompt sent from the reply plan generation support server 13 to the generation AI server 14 in the reply plan generation process according to an embodiment of the present invention. More specifically, in step S102 of the reply plan generation process, this is a statement that instructs the reply plan generation support server 13 to the generation AI server 14 to generate a question for inputting additional information and answer candidates when causing the generation AI server 14 to regenerate a reply plan.

[0068] In this prompt, the user first clarifies to the generation AI server 14 that their role is to generate a reply proposal for an email. Next, the user enters the content of the first email received and displayed on the terminal 11. While Figure 8 shows "----(insert the content of the received email)----", in reality, the following content, for example, is inserted as the content of the first email instead of this entry. "Mr. Tanaka from B Co., Ltd. Thank you very much for your continued support. This is XX from A Co., Ltd. I apologize for the personal nature of this matter, but I am writing to inform you that I will be transferring from the CE Department to the President's Office starting April 1st. I will continue to do my best to contribute to your company's business even after the transfer. After the transfer, XX in the CE department will take over my duties. ○○'s contact details are as follows: - Email address: xxxx@aaaaaa.co.jp - Phone number: 090-xxxx-xxxx Please feel free to contact us if you have any questions. We will do our best to ensure a smooth transition and appreciate your continued support. Thank you for your cooperation. ×× A Co., Ltd.

[0069] 8 clarifies that the prompt is an instruction to generate a question to be asked to the user and answer candidates so that the reply text will be optimal when generating a reply proposal for the first email. Here, the constraints on the questions and answer candidates include that the questions must be sequential, that the answer must have two candidate options, that multiple questions are allowed but that previously asked questions should not be asked, and that if the user has provided information about answer candidates, questions using that information may be included, and that three or more answer candidates for the used questions are allowed.

[0070] In FIG. 8, "----(insert content of question already asked)----" is written, but in reality, the content of the question that has been generated up to that point is written instead of this description.

[0071] The question here is preferably a closed question rather than an open question. This eliminates ambiguity in the answer and makes it easier for the generation AI server 14 to generate reply suggestions. Here, a closed question means a question that specifies the answer options and does not allow answers using free strings of characters or other information. For example, the question "What tone do you want the reply to have?" and questions in a format that displays only the answer candidates "formal" and "casual" and does not allow answers other than these are closed questions. On the other hand, questions in a format that allows free input of answers using any text in addition to or instead of these answer candidates (open questions) are not closed questions.

[0072] By making the question a closed question, answering becomes easier. For example, buttons representing each answer candidate can be displayed, and the user can select an answer with a single operation (e.g., clicking) on ​​one of the buttons, improving user convenience.

[0073] Figure 9(a) is a screen displaying a generated reply proposal according to an embodiment of the present invention, Figure 9(b) is a screen displaying a question and answer candidates according to an embodiment of the present invention, and Figure 9(c) is a screen displayed when an additional question according to an embodiment of the present invention is generated.

[0074] More specifically, FIG. 9(a) shows a screen displayed in the reply support GUI 41 when waiting for input from the user in step S9, and includes a suggested reply. The bottom of the screen displays options for input that can be received from the user. When terminal 11 receives the input "Paste into body" via the reply support GUI 41, the process proceeds to step S16, where the displayed suggested reply is reflected as the text of a new email in the email client of terminal 11 as a reply to the first email. The destination of the new email includes the sender of the first email.

[0075] On the other hand, when terminal 11 accepts the input of "regenerate reply plan," the process proceeds to step S10, where a question is generated. In this way, the "regenerate reply plan" button functions as a question generation button, and a question addition command is input by operating this button. That is, reply plan generation support server 13 receives the question addition command sent from terminal 11, and in response to this question addition command, generates an additional question for the reply plan and an additional answer candidate for the question, and transmits the additionally generated question and answer candidate to terminal 11.

[0076] 9(b) shows a screen in step S11 in which the reply support GUI 41 displayed on the terminal 11 displays the question and answer candidates. In this example, three questions are displayed, including the initially generated question and questions generated in response to two question addition commands.

[0077] When the terminal 11 accepts the input of "Add another question" in the center of the screen, the process proceeds from step S13 to step S10, and an additional question is generated. In this way, a question addition command is input by operating the "Add another question" button. While the additional question is being generated, a message indicating that processing is in progress ("Question being generated") is displayed on the screen, as shown in FIG. 9(c). When the terminal 11 accepts the input of "Regenerate" at the bottom of the screen, the process proceeds to step S14. When the terminal 11 accepts the input of "Back" at the bottom of the screen, the process proceeds to step S9 without reflecting the input from the user regarding the answer to the question displayed on the screen (not shown). Note that the display of the options "Add another question" and "Regenerate" can be grayed out to prevent user input from being received until all input of answer candidates for the question already displayed has been received.

[0078] A specific example of a reply proposal generated in step S14 is shown below: It is assumed that the reply proposal shown in Fig. 9(a) has been generated as the initial reply proposal.

[0079] For example, if "short" is selected as the answer to the question "How long should the sentence be?", the next reply suggestion generated will be something like this: "Mr. / Ms. XX Thank you for your support. This is Tanaka from B Co., Ltd. Thank you for contacting us the other day. We have received your notice regarding Mr. / Ms. XX's transfer. I am very pleased that you will continue to work with us after the transfer. I will also endeavor to ensure a smooth transition in our new collaboration with Mr. / Ms. XX. If you have any questions, please feel free to contact us. We appreciate your continued support." As you can see, the sentences are shorter than the example in Figure 9(a).

[0080] Also, if "casual" is selected as the answer to the question "What tone would you like your reply to have?", the next reply suggestion generated will be something like this: "Mr. XX, Thank you for your continued support. This is Tanaka from B Co., Ltd. I received your email the other day. First of all, congratulations on your transfer to the President's Office! I look forward to seeing you in your new position. We would also like to thank you for your contact information, Mr. / Ms. XX. We will continue to provide our full support in order to maintain a good relationship with A Co., Ltd. If you have any questions or requests, please feel free to contact us at any time. We appreciate your continued support. I wish you the best of luck in your new stage." As you can see, the tone is more casual than the example in Figure 9(a).

[0081] 9(a), when a suggested reply is displayed, the reply support GUI 41 includes only two input buttons: one for copying the suggested reply into a new email message and one for regenerating the suggested reply. This configuration allows the user to instruct copying or regenerating a suggested reply with a single operation (e.g., a single click) without having to check the details of the GUI or perform complex operations to input information.

[0082] 9(b), when the question and answer candidates are displayed, the reply support GUI 41 displays only buttons representing the answer candidates as input means for inputting an answer. With this configuration, the user can input an answer with a single operation (for example, just one click for one question) without performing a text input operation to input the answer.

[0083] 9(b), when the question and answer candidates are displayed, the reply support GUI 41 includes a button for adding a question. Furthermore, the user can easily add any number of questions by operating the button for adding questions multiple times. With this configuration, the user can add questions with a simple operation (for example, with only one click, or with any number of clicks), allowing for more flexible generation of appropriate reply suggestions.

[0084] The configuration of the input buttons can be modified in various ways, and other buttons may be included instead of or in addition to the buttons shown in FIGS. 9(a) and 9(b).

[0085] FIG. 10 is a flowchart showing the processing from when an information processing system 1 according to an embodiment of the present invention receives a summary generation command to when the summary is displayed, and explains in detail the series of processing steps S2, S5, and S6 in the flowchart shown in FIG. 3.

[0086] In step S2, terminal 11 accepts input from the user to generate a summary. Next, terminal 11 transmits a summary generation command to reply suggestion generation support server 13 (step S51). Reply suggestion generation support server 13 receives the summary generation command transmitted from terminal 11 for the first e-mail, and in response to this summary generation command, transmits a command to mail server 12 to acquire the thread (i.e., the first e-mail and the second e-mail) (step S52). Having received this command, mail server 12 transmits the first e-mail and the second e-mail to reply suggestion generation support server 13, which then receives and acquires them (step S53).

[0087] In this way, the reply suggestion generation support server 13 acquires the first email and the second email related to the first email in response to the summary generation command. In this embodiment, the reply suggestion generation support server 13 acquires both the first email and the second email, but it may also be configured to acquire only the second email.

[0088] In this embodiment, we will use the case where there are three second emails, i.e., four emails are obtained in total including the first email, as a specific example, but the actual number is not limited to this and depends on the number of emails included in the thread.

[0089] Next, the reply suggestion generation support server 13 sends to the generation AI server 14 an instruction to summarize the three second e-mails included in the thread (step S54). If there are multiple second e-mails, the reply suggestion generation support server 13 may send instructions to the generation AI server 14 in parallel to generate summaries for each second e-mail. Here, "in parallel" means that a summarization instruction for a second e-mail is sent without waiting for a response (i.e., a summary) to the summarization instruction for another second e-mail. For example, the summarization instructions for all second e-mails are sent simultaneously (or with an interval short enough for the reply suggestion generation support server 13 to process them).

[0090] By doing this, if there are multiple processing queues for the generation AI server 14 or multiple generation AI servers 14, the total time required to receive summaries of all emails from the generation AI server 14 can be shortened. Also, if the generation AI server 14 has limitations on the amount of information it can accept per command, this can be circumvented to perform the desired generation. The command sent in step S54 is explained in detail in Figure 11.

[0091] As a variant, all of the second e-mails may be combined and a single command may be issued for the combined data, or, taking into account the length of the body of the e-mail, some (for example, two) of the second e-mails may be combined and one command may be issued for the combined data and each of the second e-mails that are not combined.

[0092] In this embodiment, a case will be described in which each second e-mail contains one command, and all commands are sent in parallel to the generation AI server 14.

[0093] Next, the generation AI server 14, which has received multiple commands to generate summaries, generates summaries in accordance with the content of each command and transmits the summaries to the reply suggestion generation assistance server 13 in the order in which they are generated (steps S551 to S553). Upon receiving these summaries, the reply suggestion generation assistance server 13 generates a summary of the second email.

[0094] In this way, when generating a summary, the reply proposal generation support server 13 sends an instruction to generate a summary for the second email to the generation AI server 14, receives the summary generated by the generation AI server 14 from the generation AI server 14, and thereby generates a summary for the second email.

[0095] The reply suggestion generation support server 13 receives and generates summaries of all the second e-mails, and transmits the generated summaries to the terminal 11 (step S56). Here, the reply suggestion generation support server 13 may generate summaries of all the second e-mails based on the summaries of each second e-mail (for example, by simply combining the summaries in chronological order of the e-mails) and transmit these summaries to the terminal 11.

[0096] Terminal 11, which has received the summary, displays it on the screen of its mailer (step S6). This allows the user, when considering a reply proposal, to understand the email exchanges that have taken place so far by checking the summary, without having to check all of the related emails. As a variant, in step S54, reply proposal generation assistance server 13 may also send a command to generation AI server 14 to summarize the first email. In this case, when generating a summary, reply proposal generation assistance server 13 sends a command to generation AI server 14 to generate summaries for the first email and the second email, and receives and generates summaries of the first email and the second email from generation AI server 14.

[0097] 11 shows a prompt sent from the reply plan generation support server 13 to the generation AI server in the summary generation process according to an embodiment of the present invention. More specifically, this is a statement sent from the reply plan generation support server 13 to the generation AI server 14 in step S54 to instruct the generation AI server 14 to generate a summary.

[0098] The prompt first clarifies to the generation AI server 14 that its role is to generate a proposed reply for an email. Next, it describes the email for which the summary is to be generated, clarifying that it is an instruction to generate a summary of that email. Next, it describes the constraints suitable for generating a summary of the thread. Specific constraints include, for example, the format of the summary, not including titles or signatures unrelated to the content of the email, greetings at the beginning of the email body, and closing remarks at the end of the email body.

[0099] The above-mentioned Figure 7 can also be used as a prompt to be sent from the reply plan generation support server 13 to the generation AI server 14 in the reply plan generation process after the summary generation process according to an embodiment of the present invention. More specifically, this is a sentence sent from the reply plan generation support server 13 to the generation AI server 14 in step S3 via step S7 to instruct the generation AI server 14 to generate a reply plan.

[0100] 12 shows the screen when the generated summary is displayed on the reply support GUI 41 in step S6. Input options to be accepted from the user are displayed at the bottom of the screen, and if an input for "Generate reply proposal" is accepted, the process proceeds to step S3, and if an input for "Generate question" is accepted, the process proceeds to step S10.

[0101] As described above, in this embodiment, the mail server 12, reply proposal generation support server 13, and generation AI server 14 work together to generate reply proposals for the first email, which makes it possible to generate reply proposals for the first email efficiently and effectively, thereby achieving the significant technical effect of improving business efficiency.

[0102] The above-described system and information processing device are provided with components necessary for processing as an information processing device, such as a CPU, storage devices (including volatile and non-volatile devices), input / output devices (including a display, microphone, keyboard, mouse, speaker, and camera), even in areas not shown in the drawings, and the processing by each unit is performed by the respective hardware in association with each other. Furthermore, each process can be realized by hardware as a processing unit that performs the process, for example, by being integrated into an integrated circuit.

[0103] In the above-described embodiment, the mail server 12, reply proposal generation support server 13, and generation AI server 14 cooperate to generate a reply proposal for the first email, but a program that realizes such processing may be installed in a storage device via the Internet, etc. Furthermore, the program may be stored in advance in a storage device within the information processing device or in an external storage device, or may be installed in an external storage device when needed via a removable storage medium or communication medium (a network such as a wired, wireless, or optical network, or a carrier wave or digital signal on such a network).

[0104] In the embodiment described above, the information processing system 1 includes a terminal 11, a mail server 12, a reply suggestion generation assistance server 13, and a generation AI server 14. As a variation, the information processing system may include the reply suggestion generation assistance server 13 but may not include other computers (i.e., one or more of the terminal 11, mail server 12, and generation AI server 14). In that case, the information processing system may realize the functions described in this specification by communicating with these computers, or may operate by incorporating the functions of these computers (e.g., the mail server 12 or the generation AI server 14).

[0105] In the embodiment described above, the reply proposal generation support server 13 is an information processing device for generating summaries and reply proposals for e-mails, but it is also possible to configure it so that it only generates summaries without generating reply proposals, or conversely, it is also possible to configure it so that it only generates reply proposals without generating summaries.

[0106] Furthermore, the above-described embodiment is one example for realizing the present invention, and depending on the configuration of the device to which the present invention is applied and various conditions, it can be appropriately modified or changed within the scope of the spirit of the present invention, and the present invention is not limited to the above-described embodiment. [Explanation of symbols]

[0107] 1. Information Processing Systems 11 Terminals 12 Mail Server 13 Reply proposal generation support server 14 Generative AI Server 21 Control Unit 22 Storage device 23 Communication equipment 24 Input Devices 25 Output Devices 41 Reply support GUI NW Network

Claims

1. receiving a reply proposal generation command sent from the personal terminal in relation to the first email; acquiring the first email and a second email related to the first email; generating a reply proposal to the first email based on the first email and the second email; Sending the reply proposal to the personal device An information processing system comprising:

2. 2. An information processing system according to claim 1, wherein the reply plan generation command is input by operating a reply plan generation button displayed on the personal terminal.

3. 2. The information processing system according to claim 1, When generating the reply proposal, the information processing system - sending an instruction to a generation AI server to generate a reply proposal for said first email and said second email; - receiving from the generating AI server the reply proposal generated by the generating AI server; generating the proposed reply by An information processing system comprising:

4. 2. The information processing system according to claim 1, further comprising: receiving a question generation command transmitted from the personal terminal; generating a question related to the proposed reply and a candidate answer to the question in response to the question generation command; Sending the question and the answer candidates to the personal terminal An information processing system comprising:

5. 5. An information processing system according to claim 4, wherein the question generation command is input by operating a question generation button displayed on the personal terminal.

6. 5. The information processing system according to claim 4, further comprising: receiving a question addition command transmitted from the personal terminal; In response to the question addition command, a question for the reply proposal and a candidate answer for the question are additionally generated; The additionally generated question and answer candidates are transmitted to the personal terminal. An information processing system comprising:

7. 5. The information processing system according to claim 4, wherein the question is a closed question.

8. 2. The information processing system according to claim 1, wherein the second email is identified based on header information of the first email.

9. receiving, by the computer, a reply generation command sent from the personal terminal in relation to the first email; A step in which a computer acquires the first email and a second email related to the first email; generating a reply proposal to the first email based on the first email and the second email; a computer sending the proposed reply to the personal terminal; An information processing method comprising:

10. An information processing program that causes a computer to function as the information processing system according to claim 1.

Images