Program, information processing device, information processing method, and model generation method

Abstract

To provide a program or the like that can favorably assist in creating e-mails.SOLUTION: A program causes a computer to execute a process of acquiring first mail information including a sender and a recipient that constitutes an email, and generating second mail information by inputting the acquired first mail information into a model that generates second mail information other than the first mail information that constitutes an email when the first mail information is input. Preferably, the first mail information includes the sender and recipient and the body of the email, and inputting the first mail information into the first model generates a subject of the email.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a program, an information processing device, an information processing method, and a model generation method. [Background technology]

[0002] There is a technology that uses AI (Artificial Intelligence) to assist in creating emails, replying to them, etc. For example, Patent Document 1 discloses an email analysis server that builds a machine learning model by analyzing received emails that have already been replied to, and uses the machine learning model to estimate whether a reply to the received email is necessary. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-197851 Summary of the Invention [Problem to be solved by the invention]

[0004] In one aspect, an object of the present invention is to provide a program or the like that can favorably assist in creating e-mails. [Means for solving the problem]

[0005] In one aspect, the program causes a computer to execute a process of acquiring first email information including a sender and a recipient that constitutes an email, and generating the second email information by inputting the acquired first email information into a model that generates second email information other than the first email information that constitutes an email when the first email information is input. [Effects of the Invention]

[0006] In one aspect, it is possible to provide suitable support for creating e-mails. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is an explanatory diagram illustrating an example of the configuration of a mail generation system. [Figure 2] FIG. 2 is a block diagram illustrating an example of the configuration of a server. [Figure 3] FIG. 10 is a diagram illustrating an example of a record layout of a user DB. [Figure 4] FIG. 10 is an explanatory diagram of a subject generation model. [Figure 5] FIG. 10 is a diagram illustrating an example of a prompt to be input into a subject line generation model. [Figure 6] FIG. 1 is an explanatory diagram of a text generation model. [Figure 7] FIG. 10 is an explanatory diagram of a reply message generation model. [Figure 8] 10 is a flowchart showing the procedure of a model generation process. [Figure 9] 10 is a flowchart showing a procedure for generating second mail information. DETAILED DESCRIPTION OF THE INVENTION

[0008] The present invention will be described in detail below with reference to the drawings showing embodiments thereof. (Embodiment) FIG. 1 is an explanatory diagram showing an example of the configuration of an email generation system. In this embodiment, an email generation system that supports the creation of emails by using a machine learning model to generate the remaining information (second email information) based on a portion of the information in an email (first email information). The email generation system includes an information processing device 1, terminals 2, 2, 2..., and an email server 3. Each device is communicatively connected via a network N such as the Internet.

[0009] The information processing device 1 is an information processing device capable of various information processing and sending and receiving information, such as a server computer or a personal computer. In this embodiment, the information processing device 1 is assumed to be a server computer, and for simplicity, will be referred to as server 1 below. The server 1 performs a process of generating second mail information other than the first mail information that constitutes an e-mail, based on first mail information that constitutes an e-mail and includes at least a sender and a recipient. Specifically, as described below, the server 1 generates the subject and body of a newly created e-mail and the body (reply message) of a reply e-mail using a subject generation model 51 (see FIG. 4), a message body generation model 52 (see FIG. 6), and a reply message generation model 53 (see FIG. 7).

[0010] The terminal 2 is a terminal device used by each user, such as a personal computer, a smartphone, a tablet terminal, etc. When the server 1 receives a request to generate second mail information from a user via the terminal 2, the server 1 generates the second mail information using each model and outputs it to the terminal 2.

[0011] The mail server 3 is a server computer that stores emails sent and received by users. For example, the server 1 acquires a group of training emails from the mail server 3 and learns from them to generate a subject generation model 51, a body generation model 52, and a reply message generation model 53.

[0012] In this embodiment, the server 1 that generates the second mail information and the mail server 3 that sends and receives e-mails are described as separate computers, but they may be integrated into one computer.

[0013] 2 is a block diagram showing an example of the configuration of the server 1. The server 1 includes a control unit 11, a main memory unit 12, a communication unit 13, and an auxiliary memory unit . The control unit 11 has one or more arithmetic processing devices such as a central processing unit (CPU), a micro-processing unit (MPU), a graphics processing unit (GPU), etc., and performs various information processing, control processing, etc. by reading and executing programs stored in the auxiliary storage unit 14. The main storage unit 12 is a temporary storage area such as a static random access memory (SRAM) or a dynamic random access memory (DRAM), and temporarily stores data necessary for the control unit 11 to execute arithmetic processing. The communication unit 13 is a communication module for performing processing related to communication, and transmits and receives information to and from the outside.

[0014] The auxiliary storage unit 14 is a non-volatile storage area such as a large-capacity memory or a hard disk, and stores programs (program products) and other data required for the control unit 11 to execute processing. The auxiliary storage unit 14 also stores a subject generation model 51, a body generation model 52, a reply message generation model 53, and a user DB 141. The subject generation model 51 is a machine learning model that has trained with predetermined training data, and is a model that generates the subject of an email as second email information when the sender, destination, date and time, body, etc. of an email are input as first email information. The body generation model 52 is a machine learning model that has trained with predetermined training data, and is a model that generates the body of an email as second email information when the sender, destination, date and time, subject, etc. of an email are input as first email information. The reply message generation model 53 is a machine learning model that has learned predetermined training data, and when the sender, destination, date and time, subject, and the body of one or more emails sent and received between the sender and destination person of an email are input as first email information, the model generates the body of a reply email (reply message) as second email information. The user DB 141 is a database that stores information about users of this system.

[0015] The auxiliary storage unit 14 may be an external storage device connected to the server 1. The server 1 may be a multi-computer consisting of multiple computers, or may be a virtual machine virtually constructed by software.

[0016] Furthermore, in this embodiment, the server 1 is not limited to the above configuration, and may include, for example, an input unit that accepts operation input, a display unit that displays images, etc. Furthermore, the server 1 may be provided with a reading unit that reads a portable storage medium 1a such as a CD (Compact Disk)-ROM or a DVD (Digital Versatile Disc)-ROM, and may read and execute a program from the portable storage medium 1a.

[0017] 3 is a diagram showing an example of a record layout of user DB 141. User DB 141 includes an email address column, a name column, a department column, a job title column, and an office column. The email address column stores the email address of each user. The name column, department column, job title column, and office column each store the user's name, department name, job title, and office (work location name) in association with the email address.

[0018] Fig. 4 is an explanatory diagram of the subject generation model 51. Fig. 4 illustrates how the subject of an email is generated when information such as the sender, destination, date and time, and text of an email (newly created email) is input into the subject generation model 51.

[0019] The subject generation model 51 (and the text generation model 52 and reply message generation model 53 described below) are machine learning models that have learned predetermined training data, and when first email information that constitutes an email and includes at least a sender and a recipient is input, the model generates second email information that constitutes an email other than the first email information. For example, the subject generation model 51 is a large language model (LLM) such as a Generative Pre-trained Transformer (GPT) or a Bidirectional Encoder Representations from Transformer (BERT).

[0020] In this embodiment, the subject generation model 51, the text generation model 52, and the reply message generation model 53 are described as large-scale language models, but each model may be a model other than a large-scale language model (for example, LSTM (Long-Short Term Memory)).

[0021] In addition, in this embodiment, the subject generation model 51, the text generation model 52, and the reply message generation model 53 are described as being separate models, but as described below, the subject generation model 51, the text generation model 52, and the reply message generation model 53 may be the same model.

[0022] When subject generation model 51 receives input of primary mail information other than the subject to be generated, it generates the subject of an email as secondary mail information. Specifically, subject generation model 51 receives input of the sender, destination, current date and time, body (including the signature at the end of the body), etc. as the primary mail information.

[0023] The left side of Figure 4 illustrates an email creation screen. A user creates an email by entering recipients, CC, body, etc. on this screen. For example, terminal 2 displays a subject creation button 41 (object) in the subject entry field on this screen. When terminal 2 receives an input to create a subject, it sends to server 1 the recipients, CC, and body entered by the user, as well as the email address of the user who is the sender, and the current date and time when the subject creation button 41 was operated, and requests the creation of a subject.

[0024] When a request to generate a subject is received, the server 1 generates a subject for the email by inputting the first email information, such as the sender, destination, date and time, and text body, acquired from the terminal 2, into the subject generation model 51. Specifically, the server 1 generates a subject by creating a prompt that includes the first email information and instructs the generation of second email information and inputting this into the subject generation model 51.

[0025] FIG. 5 is a diagram showing an example of a prompt input to the subject generation model 51. For example, the prompt includes a fixed phrase that instructs the generation of a subject, a subject generation rule (constraint condition), and the first email information (date and time, sender, destination, CC, body, etc.) that is the input text. The server 1 creates the prompt by adding the fixed phrase and the generation rule to the first email information acquired from the terminal 2. When the prompt is input, the subject generation model 51 generates the subject of the email in accordance with the generation rule.

[0026] The server 1 outputs the generated subject to the terminal 2. In this case, as shown on the right side of Fig. 4, the terminal 2 automatically displays the generated subject in the subject input field. In this way, by generating a subject from information other than the subject input by the user (first mail information), it is possible to assist in creating an e-mail.

[0027] Fig. 6 is an explanatory diagram of the text generation model 52. Fig. 6 illustrates how the text of an email is generated when the sender, destination, date and time, subject, etc. of the email are input into the text generation model 52.

[0028] Like the subject generation model 51, the body generation model 52 is a model that generates second email information when it receives input of first email information, and is, for example, a large-scale language model such as GPT. When it receives input of the sender, destination, current date and time, subject, etc. of an email (newly created email) as the first email information, the body generation model 52 generates the body of the email as the second email information.

[0029] For example, terminal 2 displays a body text generation button 61 alongside an email send button on an email creation screen, as shown on the left side of Fig. 6. When terminal 2 receives an input to generate body text button 61, terminal 2 transmits to server 1 the email address of the user who is the sender, the current date and time when generate body text button 61 is operated, in addition to the destination, subject, etc. input by the user, and requests generation of the body text.

[0030] When a request to generate a text body is accepted, the server 1 generates the text body of an email by inputting the first email information, such as the sender, recipient, date and time, and subject, acquired from the terminal 2, into the text body generation model 52. Specifically, the server 1 creates a prompt that includes the first email information and instructs the generation of the text body, just as when generating the subject, and inputs this into the text body generation model 52. Although an example of the prompt is omitted here, for example, the server 1 creates a prompt that includes a fixed phrase that instructs the generation of the text body, a generation rule for the text body, and the first email information, which is the input text. The server 1 generates the text body by inputting the created prompt into the text body generation model 52 and outputs it to the terminal 2. As a result, the generated text body is automatically displayed in the text body input field, as shown on the right side of Figure 6.

[0031] Fig. 7 is an explanatory diagram of the reply message generation model 53. Fig. 7 illustrates how a reply message (the body of a reply email) is generated when the sender, recipient, date and time, subject, and body of an email sent and received between the sender and recipient are input into the reply message generation model 53.

[0032] Like the subject generation model 51 and the body generation model 52, the reply message generation model 53 is a model that generates second email information when it receives input of first email information, and is, for example, a large-scale language model such as GPT. When it receives input of the sender, destination, current date and time, subject, and body of the email sent or received up to the last time of the email (reply email) as first email information, the reply message generation model 53 generates a reply message as second email information.

[0033] For example, as shown on the left side of Fig. 7, terminal 2 displays a reply message generation button 71 on the display screen of the received email alongside a reply button and a forward button for the email received by the user (an email with the subject "Patent Application" in Fig. 7). When terminal 2 receives an operation input to the reply message generation button 71, terminal 2 transmits to server 1 the sender, destination, subject, body, etc. contained in the received email, as well as the current date and time when reply message generation button 71 was operated, and requests the generation of a reply message.

[0034] When a request to generate a reply message is accepted, the server 1 generates a reply message by inputting the first email information, such as the sender, destination, date and time, subject, and body, acquired from the terminal 2, into the reply message generation model 53. Specifically, the server 1 creates a prompt that includes the first email information and instructs the generation of a reply message, just as when generating a subject, and inputs this into the reply message generation model 53. For example, the server 1 creates a prompt that includes a fixed phrase that instructs the generation of a reply message, a reply message generation rule, and the first email information, which is the input message. The server 1 generates a reply message by inputting the created prompt into the reply message generation model 53 and outputs it to the terminal 2. As a result, as shown on the right side of Figure 7, the terminal 2 transitions to a reply message creation screen, and the reply email including the generated reply message is automatically displayed.

[0035] In this way, the server 1 generates the second email information by inputting the first email information other than the second email information to be generated (subject, body, reply) into the subject generation model 51, the body generation model 52, or the reply generation model 53. In this case, the server 1 may obtain not only the first email information such as the sender and the recipient, but also information on the person corresponding to the sender and / or the recipient, and use this information as input for each model.

[0036] The information on the person corresponding to the sender and the recipient is, for example, the name of the person corresponding to the sender and the recipient, the name of the department to which the person belongs, the job title, the name of the place of work (the name of the place of work), etc. When generating (learning) each model as described below, the server 1 learns the information on the person corresponding to the sender and / or the recipient of the email to be learned.

[0037] For example, the server 1 associates information about each user (person) with the email address of the user and stores the information about the user in the user DB 141. When the server 1 receives a request to generate second mail information from the terminal 2, if information about the person corresponding to the sender and / or destination is stored in the user DB 141, the server 1 acquires the information about the person from the user DB 141.

[0038] The server 1 generates the second email information by inputting person information acquired from the user DB 141 into each model in addition to the first email information. Specifically, the server 1 adds information about the person corresponding to the sender and / or recipient to the above-mentioned prompt and inputs it into each model. This makes it possible to change the expression of the generated sentence taking into account, for example, the relationship (hierarchical relationship) between the sender and recipient.

[0039] The server 1 may use other information as input to each model in addition to the first email information, the name corresponding to the sender and / or recipient, the department, etc. Specifically, the server 1 may use schedule information of a person corresponding to the sender and / or recipient as input. For example, when the server 1 receives a request to generate second email information, it calls a predetermined API (Application Programming Interface) and acquires schedule information from a scheduler that manages each user's schedule. The server 1 generates the second email information by creating a prompt including the acquired schedule information and the first email information, etc., and inputting this into each model. By using the schedule information as input, it is possible to preferably suggest candidate dates when creating an email for schedule adjustment, for example, as shown in FIG. 6.

[0040] The server 1 may also use information about an attachment file attached to an email (for example, the name of the attachment file) as input for each model. This allows the attachment file name to be included in the text when generating the body of a new email or a reply email.

[0041] Furthermore, the server 1 may use not only the current date and time (when the second email information is generated) but also the date and time of the last email transmission and reception between the sender and the recipient as input to each model. For example, when generating the body of a newly created email, the server 1 searches the user's inbox folder and obtains the date and time of the last email transmission and reception between the sender and the recipient. The server 1 then inputs the obtained date and time of the last email transmission and reception into the body generation model 52 to generate the body of the email. This allows the use of suitable expressions, such as changing the greeting "Thank you for your help" to "It's been a while."

[0042] Finally, the processing contents when generating the subject line generation model 51, the text body generation model 52, and the reply message generation model 53 by learning the training data will be explained.

[0043] The server 1 acquires, as training data for generating each model, a group of emails sent and received by each user from the mail server 3. The server 1 also acquires, as training data, information (such as name and department) of people corresponding to the senders and / or recipients of the acquired emails from the user DB 141.

[0044] The server 1 performs pre-learning based on training data and generates a machine learning model (large-scale language model) that generates a response sentence when an input sentence is input. That is, the server 1 inputs one or more words that make up a training sentence (such as the body of an email) into the machine learning model, predicts the word that will appear next after the one or more words, and updates parameters such as the weights between neurons so that the predicted word matches the correct word. The server 1 sequentially feeds a group of training sentences to the machine learning model to perform learning, and ultimately generates a machine learning model with optimized parameters.

[0045] Next, the server 1 performs additional learning for each task (subject generation task, body generation task, and reply message generation task) to be executed by the model, thereby generating a subject generation model 51, a body message generation model 52, and a reply message generation model 53. For example, the server 1 performs few-shot learning for each task. That is, the server 1 provides the model with a template for specifying the task and several example sentences, and causes it to perform learning.

[0046] For example, when generating subject line generation model 51, server 1 provides a fixed phrase instructing the generation of a subject line and several example subject sentences to the machine learning model to generate subject line generation model 51. When generating body line generation model 52, server 1 provides a fixed phrase instructing the generation of a body line and several example body sentences to the machine learning model to generate body line generation model 52. When generating reply line generation model 53, server 1 provides a fixed phrase instructing the generation of a reply line and several example reply sentences to the machine learning model to generate reply line generation model 53.

[0047] In this embodiment, the subject generation model 51, the body generation model 52, and the reply message generation model 53 are generated as separate models, but this embodiment is not limited to this. By training the subject generation task, the body generation task, and the reply message generation task in the same model, it is also possible to construct the same model that can execute the three tasks. In other words, the subject generation model 51 (first model), the body generation model 52 (second model), and the reply message generation model 53 (third model) may be separate models or the same model.

[0048] The server 1 generates second mail information, that is, the subject, body, and reply message of an e-mail, using the subject line generation model 51, body line generation model 52, and reply message generation model 53 generated as described above.

[0049] As described above, according to this embodiment, by inputting the first mail information, including the sender and the recipient, that constitutes the email into each model, the second mail information that constitutes the email, other than the first mail information, is generated. This can favorably support the creation of email. In particular, since the first mail information includes the sender and the recipient, the body of the email, etc., can be generated taking into consideration the relationship between the sender and the recipient.

[0050] 8 is a flowchart showing the steps of the model generation process. The process of generating a subject line generation model 51, a body generation model 52, and a reply message generation model 53 by machine learning will be described with reference to FIG. The control unit 11 of the server 1 acquires a group of emails sent and received by each user from the mail server 3 as training data (step S11). The control unit 11 also acquires information on a person (user) corresponding to the sender and / or destination of each acquired email from the user DB 141 as training data (step S12). The information on the person includes the name, department name, job title, and place of work (name of workplace).

[0051] The control unit 11 performs pre-learning based on the training data and generates a machine learning model that generates a response sentence when an input sentence is input (step S13). That is, the control unit 11 predicts the next word to appear by inputting one or more words that make up the training sentence into the machine learning model, and generates the machine learning model by optimizing parameters such as the weights between neurons so that the predicted word matches the correct word.

[0052] The control unit 11 performs additional learning of a subject generation task on the model generated in step S13, thereby generating a subject generation model 51 that generates a subject of an email (a newly created email) as second email information when the sender, destination, date and time, body, etc. of an email are input as first email information (step S14). For example, the control unit 11 generates the subject generation model 51 by performing few-shot learning, in which a fixed phrase instructing the machine learning model to generate a subject and several example subject sentences are given to the machine learning model.

[0053] The control unit 11 causes the model generated in step S13 to perform additional learning of a text generation task, thereby generating a text generation model 52 that generates the text of an email (a newly created email) as second email information when the sender, destination, date and time, subject, etc. of the email are input as first email information (step S15). For example, the control unit 11 generates the text generation model 52 by performing few-shot learning, in which a fixed phrase instructing the generation of the text and several example texts of the text are given to the machine learning model.

[0054] The control unit 11 causes the model generated in step S13 to perform additional learning of a reply message generation task, thereby generating a reply message generation model 53 that generates a reply message (the body of the reply email) as second email information when the sender, destination, date and time, subject, history of the body, etc. of an email are input as first email information (step S16). For example, the control unit 11 generates the reply message generation model 53 by performing few-shot learning, in which a fixed phrase instructing the machine learning model to generate a reply message and several example reply messages are provided. The control unit 11 then ends the series of processes.

[0055] 9 is a flowchart showing the procedure for generating second email information. The process for generating second email information using the subject generation model 51, the body generation model 52, or the reply message generation model 53 will be described with reference to FIG. The control unit 11 of the server 1 receives a request to generate second mail information from the terminal 2 (step S31). For example, the terminal 2 displays a subject generation button 41, a body generation button 61, and a reply message generation button 71 on an email creation screen or a received mail display screen, and when it receives an operation input to each button (object), it sends a request to the server 1 to generate a subject, a body, or a reply message.

[0056] The control unit 11 acquires first mail information including at least the sender and the recipient that constitute the email from the terminal 2 (step S32). Specifically, when a request to create a subject is received, the control unit 11 acquires the sender, recipient, date and time, body, etc. of the email as the first mail information. When a request to create a body is received, the server 1 acquires the sender, recipient, date and time, subject, etc. of the email as the first mail information. When a request to create a reply message is received, the server 1 acquires the sender, recipient, date and time, subject, body of the email sent and received between the sender and recipient (body of the received email), etc. as the first mail information. In addition, the control unit 11 may acquire information about an attachment attached to the email (such as the name of the attachment).

[0057] The control unit 11 acquires information about a person corresponding to the sender and / or destination included in the acquired first mail information from the user DB 141 (step S33). The information about the person includes the name, department name, job title, and place of work (name of workplace).

[0058] The control unit 11 determines whether a request for generating a subject has been received in step S31 (step S34). If it is determined that a request for generating a subject has been received (S34: YES), the control unit 11 generates a subject of the email (newly created email) as second email information by inputting the first email information acquired in step S32 and the person information acquired in step S33 into the subject generation model 51 (step S35). Specifically, the control unit 11 creates a prompt including a fixed phrase instructing the generation of a subject, a subject generation rule, the first email information, and information (input sentence) of a person corresponding to the sender and / or recipient, and inputs the prompt into the subject generation model 51 to generate a subject of the email. The control unit 11 proceeds to step S39.

[0059] If it is determined that a request for generating a subject has not been received (S34: NO), the control unit 11 determines whether a request for generating a body has been received (step S36). If it is determined that a request for generating a body has been received (S36: YES), the control unit 11 generates the body of the email (newly created email) as second email information by inputting the first email information acquired in step S32 and the person information acquired in step S33 into the body generation model 52 (step S37). Specifically, the control unit 11 creates a prompt including a fixed phrase instructing the generation of the body, a generation rule for the body, the first email information, and information (input sentence) of the person corresponding to the sender and / or recipient, and inputs the prompt into the body generation model 52 to generate the body of the email. The control unit 11 proceeds to step S39.

[0060] If it is determined that a request to generate a text body has not been received (S36: NO), that is, if a request to generate a reply message has been received, the control unit 11 generates a reply message (the text body of the reply email) as second mail information by inputting the first mail information acquired in step S32 and the person information acquired in step S33 into the reply message generation model 53 (step S38). Specifically, the control unit 11 creates a prompt including a fixed phrase instructing the generation of a reply message, a reply message generation rule, the first mail information, and information (input text) of the person corresponding to the sender and / or recipient, and inputs the prompt into the reply message generation model 53 to generate a reply message. The control unit 11 proceeds to step S39.

[0061] The control unit 11 outputs the subject generated in step S35, the body generated in step S37, or the reply message generated in step S38 to the terminal 2 (step S39), and ends the series of processes.

[0062] In the above description, the machine learning models (subject generation model 51, body generation model 52, and reply message generation model 53) themselves are made to learn information (such as name and department) of people corresponding to the sender and / or recipient, but this embodiment is not limited to this. The server 1 may simply input information about people corresponding to the sender and / or recipient into the models when generating the second email information, without having to have the models learn information about the people. Simply providing information about the people when generating the second email information makes it possible to generate a sentence that takes into account the relationship between the two parties, etc.

[0063] As described above, according to this embodiment, it is possible to provide suitable support for creating e-mails.

[0064] The embodiments disclosed herein are to be considered as illustrative in all respects and not restrictive. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims.

[0065] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any combination, regardless of the reference format. Furthermore, although the claims use a format in which a claim references two or more other claims (multi-claim format), this is not limited to this format. A format in which multiple claims (multi-multi-claims) that reference at least one other multiple claim may also be used. [Explanation of symbols]

[0066] 1. Server (information processing device) 11 Control section 12 Main memory 13 Communications Department 14 Auxiliary storage 51 Subject Generation Model (First Model) 52 Text generation model (2nd model) 53 Reply generation model (third model) 141 User DB 2. Devices 3. Mail Server

Claims

1. Acquire first mail information including a sender and a recipient, which constitutes an email; When the first mail information is input, the acquired first mail information is input into a model that generates second mail information other than the first mail information, which constitutes an email, thereby generating the second mail information. A program that causes a computer to perform a process.

2. the first mail information includes the sender and destination and the body of the e-mail; Generate an email subject by inputting the first email information into a first model. The program according to claim 1.

3. The first email information includes the sender and destination, the date and time, and the subject of the email, The first email information is input into a second model to generate the body of an email. The program according to claim 1.

4. The first email information includes the sender and the recipient, the date and time, the subject of the email, and the body of the email sent and received between the sender and the recipient, The first email information is input into a third model to generate a reply email body for the sent or received email. The program according to claim 1.

5. Acquire information about a person corresponding to the sender or recipient from a database that stores information about the person corresponding to the email address in association with the email address; The acquired information about the person and the first email information are input into the model to generate the second email information. The program according to claim 1.

6. The person's information includes the person's name, department name, job title, or place of work. The program according to claim 5.

7. Obtain schedule information of a person corresponding to the sender or recipient; The second email information is generated by inputting the schedule information and the first email information into the model. The program according to claim 1.

8. On an email creation screen, an object for generating the second email information is displayed in association with an input field for inputting the second email information; When an operation input to the object is received, the second mail information is generated; The second email information is displayed in the input field. The program according to claim 1.

9. displaying an object for generating the second email information on a display screen of the received email; When an operation input to the object is received, the second mail information is generated; The screen transitions to a screen for creating a reply email to the received email, and the reply email including the second email information is displayed. The program according to claim 1.

10. An information processing device including a control unit, The control unit Acquire first mail information including a sender and a recipient, which constitutes an email; When the first mail information is input, the acquired first mail information is input into a model that generates second mail information other than the first mail information, which constitutes an email, thereby generating the second mail information. Information processing device.

11. Acquire first mail information including a sender and a recipient, which constitutes an email; When the first mail information is input, the acquired first mail information is input into a model that generates second mail information other than the first mail information, which constitutes an email, thereby generating the second mail information. An information processing method in which processing is performed by a computer.

12. Get a batch of emails, Based on the group of emails, when first email information including a sender and a recipient that constitutes an email is input, a model is generated that generates second email information other than the first email information that constitutes an email. A model generation method in which processing is performed by a computer.

13. Acquire information about a person corresponding to the sender or destination of each acquired email from a database that stores information about the person corresponding to the email address in association with the email address; The model is generated based on the acquired information about the person and the email group. The model generation method of claim 12.

Images