Schedule adjustment system, schedule adjustment method and program

Abstract

To provide a schedule adjustment system for improving convenience of users.SOLUTION: A communication data acquisition unit (201) of a schedule adjustment system (1) acquires communication data representing content of communication performed with a communication tool. An adjustment schedule data acquisition unit (203) acquires adjustment schedule data representing an adjustment schedule to be adjusted through communication on the basis of communication data and a machine learning model capable of executing language analysis. A schedule transmission unit (206) transmits the adjustment schedule data to a schedule management tool cooperating with the communication tool.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present disclosure relates to a schedule adjustment system, a schedule adjustment method, and a program.

Background Art

[0002] Conventionally, communication tools for users to communicate by email or the like are known. For example, a user may use a communication tool to adjust schedules for meetings or the like. When a schedule is finalized, the user registers the schedule in a schedule management tool. When schedule adjustment and registration are performed in such a process, the user has to go back and forth between the communication tool and the schedule management tool, so the user's operation is cumbersome.

[0003] For example, Non-Patent Document 1 describes email software which is an example of a communication tool. The email software of Non-Patent Document 1 acquires schedule data indicating a schedule such as a flight or accommodation from an email sent by a specific sender such as an airline or a lodging facility. The email software of Non-Patent Document 1 automatically adds the schedule indicated by the schedule data acquired from the email to the schedule table in the email software.

Prior Art Documents

Non-Patent Documents

[0004]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, the email software of Non-Patent Document 1 can only handle emails with a specific format sent by a specific sender. Emails from the person with whom the user adjusts the schedule, such as emails from airlines or accommodation facilities, do not have a defined format, so the email software of Non-Patent Document 1 cannot acquire schedule data. For this reason, the email software of Non-Patent Document 1 lacks versatility and cannot sufficiently enhance the convenience of the user.

[0006] One of the objectives of the present disclosure is to enhance the convenience of the user.

Means for Solving the Problems

[0007] A schedule adjustment system according to an aspect of the present disclosure includes a communication data acquisition unit that acquires communication data indicating the content of communication performed by a communication tool, an adjustment schedule data acquisition unit that acquires adjustment schedule data indicating an adjustment schedule that is a schedule adjusted by the communication, based on the communication data and a machine learning model capable of language analysis, and a schedule transmission unit that transmits the adjustment schedule data to a schedule management tool that cooperates with the communication tool.

Advantages of the Invention

[0008] According to the present disclosure, the convenience of the user can be enhanced.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Embodiments for Carrying Out the Invention

[0010] [1. Hardware Configuration of Schedule Adjustment System] An example of an embodiment of a schedule adjustment system, a schedule adjustment method, and a program according to the present disclosure will be described. FIG. 1 is a diagram showing an example of the hardware configuration of a schedule adjustment system. For example, the schedule adjustment system 1 includes a learning terminal 10, a communication server 20, a schedule management server 30, and a user terminal 40. Each of the learning terminal 10, the communication server 20, the schedule management server 30, and the user terminal 40 is connected to a network N such as the Internet or a LAN.

[0011] The learning terminal 10 is a computer that performs learning of the machine learning model described later. For example, the learning terminal 10 is a personal computer, a tablet terminal, or a smartphone. For example, the learning terminal 10 includes a control unit 11, a storage unit 12, a communication unit 13, an operation unit 14, and a display unit 15. The control unit 11 includes at least one processor. The storage unit 12 includes at least one of a volatile memory such as a RAM and a non-volatile memory such as a flash memory. The communication unit 13 includes at least one of a communication interface for wired communication and a communication interface for wireless communication. The operation unit 14 is an input device such as a mouse or a touch panel. The display unit 15 is a liquid crystal or organic EL display.

[0012] The communication server 20 is a server computer. For example, the communication server 20 includes a control unit 21, a storage unit 22, and a communication unit 23. The hardware configurations of the control unit 21, the storage unit 22, and the communication unit 23 may be the same as those of the control unit 11, the storage unit 12, and the communication unit 13, respectively.

[0013] The schedule management server 30 is a server computer. For example, the schedule management server 30 includes a control unit 31, a storage unit 32, and a communication unit 33. The hardware configurations of the control unit 31, the storage unit 32, and the communication unit 33 may be the same as those of the control unit 11, the storage unit 12, and the communication unit 13, respectively.

[0014] The user terminal 40 is the user's computer. For example, the user terminal 40 is a personal computer, a tablet terminal, or a smartphone. For example, the user terminal 40 includes a control unit 41, a storage unit 42, a communication unit 43, an operation unit 44, and a display unit 45. The hardware configurations of the control unit 41, the storage unit 42, the communication unit 43, the operation unit 44, and the display unit 45 may be the same as those of the control unit 11, the storage unit 12, the communication unit 13, the operation unit 14, and the display unit 15, respectively.

[0015] Note that the programs stored in the storage units 12, 22, 32, and 42 may be supplied via the network N. The hardware configurations of each of the learning terminal 10, the communication server 20, the schedule management server 30, and the user terminal 40 are not limited to the example in FIG. 1. For example, at least one of the learning terminal 10, the communication server 20, the schedule management server 30, and the user terminal 40 may include at least one of a reading unit (e.g., a memory card slot) for reading a computer-readable information storage medium and an input / output unit (e.g., a USB terminal) for directly connecting to an external device. A program stored in the information storage medium may be supplied to at least one of the learning terminal 10, the communication server 20, the schedule management server 30, and the user terminal 40 via at least one of the reading unit and the input / output unit.

[0016] Also, the schedule adjustment system 1 may include at least one computer. The computer included in the schedule adjustment system 1 is not limited to the example in FIG. 1. For example, the schedule adjustment system 1 may include only the learning terminal 10, the communication server 20, and the schedule management server 30. In this case, the user terminal 40 exists outside the schedule adjustment system 1. The schedule adjustment system 1 may include only the communication server 20. In this case, the learning terminal 10, the schedule management server 30, and the user terminal 40 exist outside the schedule adjustment system 1. The schedule adjustment system 1 may include the communication server 20 and another server computer other than the schedule management server 30.

[0017] [2. Outline of Schedule Adjustment System] In this embodiment, the user adjusts the schedule using a communication tool. The communication tool is a tool for the user to communicate with others. The communication tool can also be said to be a program for the user to send messages to others. The communication tool may be a known tool. For example, the communication tool may be an email tool, a chat tool, a messaging app, SMS (Short Message Service), or an SNS (Social Networking Service).

[0018] In this embodiment, as an example of the communication tool, an email tool will be described. For example, the user uses groupware having a communication tool. The groupware is a program for supporting the user's work. The groupware may be either cloud-based or on-premises. In this embodiment, the communication server 20 manages the communication tool. When the user logs in to the groupware and selects the communication tool, the user terminal 40 causes the display unit 45 to display a communication screen, which is a screen of the communication tool.

[0019] FIG. 2 and FIG. 3 are diagrams showing an example of the communication screen. In the examples of FIG. 2 and FIG. 3, a user named "Taro Ayabou" who works for a certain company uses email to communicate with a person named "Hanako Kanto" outside the company and adjusts the schedule for a meeting. The schedule may be anything that takes place at a predetermined date and time at a predetermined location. The schedule is not limited to meetings like those in FIG. 2 and FIG. 3. For example, the schedule may be a meeting outside of work, going out, a business trip, a seminar, or an interview. The schedule is not limited to those taking place in the physical space. The schedule may be something that takes place in a virtual space or online. For example, a web conference held at a predetermined URL corresponding to a location on the Internet may correspond to the schedule.

[0020] Hereinafter, the schedule to be adjusted by the user is referred to as the adjustment schedule. The person with whom the user plans to make an adjustment (the person with whom the user communicates, "Kanato Hanako" in the examples of FIGS. 2 and 3) is referred to as the adjustment partner. For example, the user sends an email to the email address of the adjustment partner, listing multiple dates and times as candidates. The adjustment partner sends an email to the user's email address, listing the date and time that they prefer among the multiple dates and times. In the example of the communication screen SC1 in the upper part of FIG. 2, the adjustment partner prefers the time "October 26, 2023, 13:00 - 14:00".

[0021] In this embodiment, the communication server 20 supports the user's adjustment task by identifying the date and time of the adjustment schedule, etc. from the email based on a machine learning model capable of language analysis. Details of the machine learning model will be described later. For example, when the user selects the button B10 on the communication screen SC1, the communication server 20 uses the machine learning model to identify the time "October 26, 2023, 13:00 - 14:00" that the adjustment partner prefers from the email. The email also includes other dates and times initially presented by the user, but in this embodiment, it is assumed that the machine learning model is trained to identify the date and time included in the relatively new reply content rather than the other dates and times included in the relatively old reply content.

[0022] For example, the communication screen SC1 switches to a reply mode, which is a mode for replying to emails, as shown in the lower part of FIG. 2. The communication screen SC1 in the reply mode includes a display area A11 indicating the date and time of the adjustment schedule identified by the machine learning model. If the date and time of the adjustment schedule identified by the machine learning model is incorrect, the user can correct the date and time of the adjustment schedule from the display area A11.

[0023] For example, when the user selects button B110 for schedule confirmation, communication server 20 checks the user's schedule based on the schedule management tool of groupware. The schedule management tool is a tool for managing the user's schedule. In this embodiment, a case where the schedule management tool is a tool of the same groupware as the communication tool will be described, but the schedule management tool may not be a tool of groupware. In this embodiment, a case where schedule management server 30 manages the schedule management tool will be taken as an example.

[0024] Hereinafter, a schedule registered in the schedule management tool will be referred to as a registered schedule. For example, communication server 20 checks whether the date and time of the adjustment schedule is available based on the date and time of the adjustment schedule and the date and time of the registered schedule. When it is confirmed that the date and time of the adjustment schedule is available, as shown in the upper part of FIG. 3, user terminal 40 causes a message indicating that the date and time of the adjustment schedule is available to be displayed in display area A11.

[0025] Note that the machine learning model may identify not only the date and time of the adjustment schedule but also at least one of the title and memo of the adjustment schedule from the email. For example, when the machine learning model can identify at least one of the title and memo of the adjustment schedule, user terminal 40 causes at least one of the title and memo of the adjustment schedule to be displayed in display area A11. In the example of communication screen SC1 in the upper part of FIG. 3, it is assumed that the machine learning model can identify a title such as "Mr. Jindou of yyy Co., Ltd." from a part such as the recipient of the email. The title is displayed in display area A11. The user can modify at least one of the title and memo of the adjustment schedule from display area A11.

[0026] For example, when the user selects button B111 for scheduled transmission, communication server 20 transmits adjustment schedule data indicating an adjustment schedule to schedule management server 30 that manages the schedule management tool based on the content displayed in display area A11. Schedule management server 30 registers the adjustment schedule in the schedule management tool based on the adjustment schedule data. Communication server 20 generates a reply text based on a predetermined template. As shown in the lower part of Figure 3, user terminal 40 displays on communication screen SC1 a message indicating that the user's schedule has been transmitted and a reply text generated based on the template. When the user selects button B12 for email transmission, an email is sent to the email address of the adjustment counterpart.

[0027] Note that the machine learning model may generate the reply text. For example, on communication screen SC1, the reply text generated by the machine learning model may be displayed. When the user selects button B112, communication server 20 may cause the machine learning model to modify the reply text. The machine learning model may modify the reply text in the same manner as a known dialogue model.

[0028] For example, the user may cause the machine learning model to modify the reply text by inputting an instruction such as "Please use a more polite expression." The process of the machine learning model modifying the text based on the user's instruction may be a known process. When the machine learning model modifies the text based on the user's instruction, user terminal 40 displays the modified text on communication screen SC1. The user may instruct the machine learning model to make another modification. When the user selects the schedule management tool from communication screen SC1, user terminal 40 displays the schedule management screen, which is the screen of the schedule management tool, on display unit 45.

[0029] FIG. 4 is a diagram showing an example of a schedule management screen. As shown in FIG. 4, the schedule management screen SC2 shows the registered schedules registered in the schedule management tool. In the example of FIG. 4, the adjustment schedule (the schedule at the time of "13:00 - 14:00 on October 26, 2023") adjusted by the user with the adjustment partner on the communication screen SC1 is registered in the schedule management tool as a registered schedule. When the user selects the registered schedule displayed on the schedule management screen SC2, the user terminal 40 causes the schedule management screen SC2 to display the details of the registered schedule.

[0030] As described above, the schedule adjustment system 1 uses a machine learning model to identify the date and time of the adjustment schedule described in the email. The schedule adjustment system 1 causes the user to correct the date and time as necessary, and then transmits adjustment schedule data indicating the identified date and time to the schedule management tool. By using a machine learning model, the schedule adjustment system 1 can identify the date and time of the adjustment schedule even if the email has no defined format, thus enhancing the convenience for the user. Hereinafter, the details of the schedule adjustment system 1 will be described.

[0031] [3. Functions Realized by the Schedule Adjustment System] FIG. 5 is a diagram showing an example of the functions realized by the schedule adjustment system 1.

[0032] [3-1. Functions Realized by the Learning Terminal] For example, the learning terminal 10 includes a data storage unit 100 and a learning unit 101. The data storage unit 100 is realized by the storage unit 12. The learning unit 101 is realized by the control unit 11.

[0033] [Data Storage Unit] The data storage unit 100 stores data necessary for training the machine learning model M. For example, the data storage unit 100 stores the machine learning model M. The machine learning model M is a model created based on machine learning techniques. The machine learning model M may be any type of model used in the field of natural language processing. For example, the machine learning model M may be a model of supervised learning, semi-supervised learning, or unsupervised learning.

[0034] In this embodiment, a case where the machine learning model M is a so-called large language model will be exemplified. For example, the machine learning model M may be a large language model such as GPT (Generative Pre-trained Transformer), BERT (Bidirectional Encoder Representations from Transformers), PaLM (Pathways Language Model), or LLaMA (Large Language Model Meta AI). The machine learning model M may be another model not classified as a large language model (for example, a neural network or a sequence-to-sequence model).

[0035] For example, the machine learning model M includes parameters that are adjusted by training. The parameters of the machine learning model M are referenced by the program of the machine learning model M. For example, the parameters are weight coefficients and biases. The parameters may be various known parameters. The parameters are not limited to weight coefficients and biases. For example, the parameters may be a matrix referenced during the calculation of embedded representations, a positional encoding referenced in the encoding of token positions, or other parameters.

[0036] The machine learning model M includes programs for calculating embedded representations and the like. The program of the machine learning model M includes code indicating the internal processing of the machine learning model M. For example, the program of the machine learning model M includes an encoder for calculating an embedded representation, a decoder for creating data for output according to tasks such as schedule adjustment, an output layer for performing a final output based on the data, and processing of other intermediate layers. When the machine learning model M is a large language model, the machine learning model M also includes a program indicating a process of dividing the input data into a plurality of tokens. The program of the machine learning model M may be a known program.

[0037] In the present embodiment, a case where a pre-trained large language model corresponds to the machine learning model M is taken as an example. For example, the learning unit 101 described later performs re-learning of the machine learning model M. Re-learning is an adjustment of the parameters of the pre-trained machine learning model M. For example, re-learning is fine-tuning, transfer learning, or distillation. The learning unit 101 described later may perform learning of the machine learning model M (the machine learning model M with initial parameter values) for which pre-learning has not been performed instead of re-learning of the machine learning model M. The data storage unit 100 stores not only the machine learning model M but also the training database DB1 necessary for learning.

[0038] FIG. 6 is a diagram showing an example of the training database DB1. The training database DB1 is a database in which training data necessary for learning the machine learning model M is stored. For example, the training data includes an input part input to the machine learning model M during learning and an output part that becomes the correct answer during learning. The combination of the input part and the output part may be any combination according to the content of the task executed by the machine learning model M. The training data may be prepared manually or by a known tool.

[0039] For example, the input part of the training data is all or part of the data indicating the training emails. The training emails may be actual sent emails or emails with virtual content. The training emails have some information related to the training schedule described therein. For example, the training emails have the date and time, title, memo, location, participants, or a combination thereof of the training schedule described therein. The training emails may also have other information such as greetings or small talk not related to the schedule described therein.

[0040] For example, the output part of the training data is the part to be specified from the training emails. The output part of the training data indicates some text of the training emails. For example, the output part of the training data is the part of the training emails where the date and time, title, memo, location, participants, or a combination thereof of the training schedule are described. The output part of the training data may be annotated by the creator of the training data or by a known tool. In the example of FIG. 6, the output part of the training data indicates the date and time of the training schedule.

[0041] For example, like the emails in FIGS. 2 and 3, there may be multiple dates and times described in the relatively old reply content and the date and time desired by the adjustment partner described in the relatively new reply content. Training data may be created so that the machine learning model M can handle such emails. For example, the input part of the training data may be a training email with multiple dates and times described in the relatively old reply content and the date and time desired by the adjustment partner described in the relatively new reply content. The output part of the training data may be the date and time included in the relatively new reply content. With such training data, the machine learning model can learn which part of the email the described date and time should be specified for.

[0042] Note that the data stored in the data storage unit 100 is not limited to the examples of the present embodiment. The data storage unit 100 only needs to store data necessary for the learning of the machine learning model M. For example, the data storage unit 100 stores a learning program showing a series of processes in the learning of the machine learning model M. Assume that the calculation formula of the loss function calculated during learning is also shown in the learning program.

[0043] [Learning Unit] The learning unit 101 executes the learning of the machine learning model M based on each of a plurality of pieces of training data stored in the training database DB1. In the present embodiment, an example in which the learning unit 101 executes the learning of the machine learning model M based on the algorithm of supervised learning is given. The learning algorithm may be a known algorithm. For example, the learning unit 101 may execute the learning of the machine learning model M based on the algorithm of semi-supervised learning or unsupervised learning.

[0044] For example, based on the algorithm of supervised learning, when the input part of the training data is input to the machine learning model M, the learning unit 101 adjusts the parameters of the machine learning model M so that the output part of the training data is output from the machine learning model M, thereby executing the learning of the machine learning model M. The learning unit 101 calculates a loss indicating the error between the output from the machine learning model M when the input part of the training data is input to the machine learning model M and the output part of the training data, based on a known loss function. The learning unit 101 executes the learning of the machine learning model M by adjusting the parameters of the machine learning model M so that the calculated loss becomes small. The learning unit 101 may repeatedly execute the learning of the machine learning model M until the loss becomes less than the threshold value.

[0045] For example, during learning, the machine learning model M divides the text indicated by the input part of the training data into a plurality of tokens. The machine learning model M encodes each of the plurality of tokens and converts them into a sequence of embedding representations. Based on the sequence of embedding representations, the machine learning model M predicts its continuation as needed and calculates an output corresponding to the task. The greater the difference between the output of the machine learning model M and the output part of the training data, the greater the loss. The smaller the difference between the output of the machine learning model M and the output part of the training data, the smaller the loss. These series of processes are executed based on the program of the machine learning model M (for example, the encoder, decoder, output layer, and other layers) and the current parameters of the machine learning model M.

[0046] Note that the algorithm used for learning may be a known algorithm. For example, the learning unit 101 executes the learning of the machine learning model M based on the backpropagation algorithm. The learning unit 101 may execute the learning of the machine learning model M based on other algorithms other than the backpropagation algorithm. For example, the learning unit 101 may execute the learning of the machine learning model M based on the gradient descent method, the momentum method, the quasi-Newton method, the conjugate gradient method, the local search method, or other algorithms.

[0047] In this embodiment, the learning unit 101 executes a series of learning processes by executing the learning program stored in the data storage unit 100. The learning unit 101 records the learned machine learning model M in the data storage unit 100. The data storage unit 100 may store both the machine learning model M before learning and the learned machine learning model M. The learning unit 101 transmits the learned machine learning model M to the communication server 20. The learned machine learning model M transmitted to the communication server 20 is provided for the user to use.

[0048] [3-2. Functions Realized by the Communication Server] For example, the communication server 20 includes a data storage unit 200, a communication data acquisition unit 201, a task determination unit 202, a scheduled adjustment data acquisition unit 203, a scheduled registration data acquisition unit 204, a display control unit 205, a scheduled transmission unit 206, and a reply content generation unit 207. The data storage unit 200 is realized by the storage unit 22. Each of the communication data acquisition unit 201, the task determination unit 202, the scheduled adjustment data acquisition unit 203, the scheduled registration data acquisition unit 204, the display control unit 205, the scheduled transmission unit 206, and the reply content generation unit 207 is realized by the control unit 21.

[0049] [Data storage unit] The data storage unit 200 stores data necessary for supporting schedule adjustment. For example, the data storage unit 200 stores a communication database DB2 and a learned machine learning model M.

[0050] FIG. 7 is a diagram showing an example of the communication database DB2. The communication database DB2 is a database in which communication data indicating the content of communication performed by the communication tool is stored. For example, the communication database DB2 stores a communication ID and communication data.

[0051] The communication ID is an ID for the schedule adjustment system 1 to identify communication data. When a new communication is performed by the communication tool, the communication server 20 generates new communication data indicating the new communication. The communication server 20 issues a communication ID for the new communication data. The communication server 20 stores the communication ID and the new communication data in the communication database DB2.

[0052] Communication data is text indicating the specific content of communication. Here, the text does not refer to data with a.txt extension, but rather means a combination of at least one of characters, numbers, and other symbols. That is, the text is a passage of a certain length. Note that the communication data may be data with a.txt extension. The communication data may be in any data format. For example, as in this embodiment, when the communication data is email data, the communication data may be data of a message file with an extension such as.msg, or it may be data of a character string indicating the text extracted from the email.

[0053] In this embodiment, an example is given where the communication data is data of a character string indicating the body of an email. The communication data may indicate parts other than the body of the email (for example, the header). The communication data may be data corresponding to the communication tool. The communication data is not limited to email data. For example, the communication data may be data indicating messages exchanged in a chat tool, a messaging app, SMS, or SNS.

[0054] Note that the data stored in the data storage unit 200 is not limited to the above examples. The data storage unit 200 can store any data. For example, the data storage unit 200 may store a user database in which various user data is stored. The user database stores data such as user IDs and passwords. The data storage unit 200 may store data of each screen such as the communication screen SC1. The data storage unit 200 may store template data indicating a template for reply content. The data storage unit 200 may store a database similar to the schedule management database DB3 described later.

[0055] [Communication Data Acquisition Unit] The communication data acquisition unit 201 acquires communication data. In the present embodiment, since the communication data is stored in the communication database DB2, the communication data acquisition unit 201 acquires the communication data from the communication database DB2. The communication data may be stored in another database other than the communication database DB2. The communication data may be stored in another computer other than the communication server 20. The communication data acquisition unit 201 may acquire the communication data from the other database or the other computer.

[0056] In the present embodiment, the communication data acquisition unit 201 acquires the communication data of the mail designated by the user among the mails addressed to the user. For example, in the state of the communication screen SC1 on the upper side of FIG. 2, when the user selects the button B10, the communication data acquisition unit 201 acquires the communication data indicating the displayed mail. The communication data acquisition unit 201 may acquire the communication data indicating the displayed mail before the user selects the button B10. The communication data acquisition unit 201 may acquire the communication data indicating the mail in which the matter to be adjusted for processing is described.

[0057] [Task determination unit] The task determination unit 202 determines whether the task of the machine learning model M is schedule adjustment based on the communication data. The task of the machine learning model M is the content of the process that the machine learning model M should execute. The machine learning model M may be a model specialized for a specific task or a model not specialized for a specific task. For example, when the machine learning model M is a highly general-purpose GPT model, the machine learning model M may not be specialized for a specific task. In this case, in addition to the task of schedule adjustment, the machine learning model M can also perform tasks such as machine translation, summary creation, data classification, or other tasks. The task that the machine learning model M should execute may be specified as an input (prompt) to the machine learning model M.

[0058] The method for determining the task of the machine learning model M may be a known method. For example, the task determination unit 202 inputs the communication data to the machine learning model M. The machine learning model M divides the text indicated by the communication data into a plurality of tokens. The machine learning model M calculates an embedding representation based on the plurality of tokens and predicts the continuation of the text as necessary. The machine learning model M analyzes the context of the text based on the calculated embedding representation to identify the task it should perform.

[0059] For example, when the machine learning model M determines that the text indicated by the communication data contains specific text related to schedule adjustment (for example, text indicating some date and time, or text used in schedule adjustment such as "How about your convenience at the following date and time") based on the embedding representation, the machine learning model M determines that its task is schedule adjustment. When the machine learning model M determines that the text indicated by the communication data does not contain specific words related to schedule adjustment based on the embedding representation, the machine learning model M determines that its task is not schedule adjustment.

[0060] Note that the task determination unit 202 may determine the task of the machine learning model M without inputting the communication data into the machine learning model M. For example, the task determination unit 202 may determine whether the task of the machine learning model M is schedule adjustment by determining whether the text indicated by the communication data contains text used for schedule adjustment, such as "meeting", based on a dictionary prepared in advance. Assume that terms indicating schedule adjustment tasks are registered in the dictionary. The task determination unit 202 may determine that the task of the machine learning model M is a schedule adjustment task when a predetermined number or more of the terms registered in the dictionary exist in the text indicated by the communication data.

[0061] For another example, the task determination unit 202 may determine the task of the machine learning model M by using another machine learning model for task determination separately from the machine learning model M for specifying the date and time of the adjustment schedule, etc. In this case, it is assumed that the relationship between the training communication data and the task of the machine learning model M is learned in another machine learning model. Another machine learning model labels whether the communication data input to itself belongs to a schedule adjustment task. The task determination unit 202 may determine that the task of the machine learning model M is schedule adjustment when the label output by another machine learning model indicates a schedule adjustment task.

[0062] [Schedule adjustment data acquisition unit] The schedule adjustment data acquisition unit 203 acquires schedule adjustment data indicating a schedule adjustment that is a schedule adjusted by communication, based on the communication data and the machine learning model M capable of language analysis. The machine learning model M capable of language analysis is a machine learning model M that can calculate an embedding representation and perform an output corresponding thereto when some text is input. The machine learning model M used in the field of natural language processing described above corresponds to the machine learning model M capable of language analysis.

[0063] The adjustment schedule data is data indicating the content of the adjustment schedule. For example, the adjustment schedule data acquisition unit 203 acquires adjustment schedule data indicating the date and time of the adjustment schedule. In this embodiment, since the situation before the adjustment schedule is finalized is taken as an example, the date and time that is a candidate for the adjustment schedule corresponds to the date and time of the adjustment schedule. On the email, the adjustment schedule may be finalized. In this case, the adjustment schedule data may indicate the finalized date and time of the adjustment schedule instead of the date and time that is a candidate for the adjustment schedule.

[0064] For example, the adjustment schedule data acquisition unit 203 may acquire adjustment schedule data indicating the date and time of the adjustment schedule and other information of the adjustment schedule. The other information may be any information other than the date and time. In this embodiment, the other information is at least one of the title of the adjustment schedule and the memo of the adjustment schedule. The other information may be the name of the participant in the adjustment schedule, the company information of the participant, the place where the adjustment schedule is held, or a facility such as a conference room used in the adjustment schedule. For an online meeting, the other information may be the URL used in the online meeting. Note that the adjustment schedule data may indicate only the other information of the adjustment schedule without indicating the date and time of the adjustment schedule. The adjustment schedule data may indicate a part of the text related to the adjustment schedule among the text indicated by the communication data.

[0065] For example, the adjustment schedule data acquisition unit 203 inputs the communication data to the machine learning model M. The input of the communication data to the machine learning model M may be performed by the task determination unit 202 described above. The machine learning model M performs acquisition of a plurality of tokens and calculation of embedding representations as described in the above flow. The machine learning model M analyzes the context of the text based on the embedding representation and identifies a part related to the adjustment schedule among the text indicated by the communication data. The machine learning model M outputs adjustment schedule data indicating the identified part. The adjustment schedule data acquisition unit 203 acquires the adjustment schedule data output from the machine learning model M.

[0066] In the present embodiment, since the task determination unit 202 executes task determination, the scheduled adjustment data acquisition unit 203 acquires scheduled adjustment data when the task determination unit 202 determines that the task is a scheduled adjustment. When the task determination unit 202 does not determine that the task is a scheduled adjustment, the scheduled adjustment data acquisition unit 203 does not acquire the scheduled adjustment data. When the task determination unit 202 does not determine that the task is a scheduled adjustment, the communication data may not be input to the machine learning model M.

[0067] In the case of the interactive machine learning model M, the scheduled adjustment data acquisition unit 203 may input, to the machine learning model M, data indicating the task to be executed by the machine learning model M together with the communication data. For example, when the user selects the button B10, the scheduled adjustment data acquisition unit 203 may input, to the machine learning model M, data indicating an instruction such as "Specify and output the scheduled adjustment date and time, title, and memo from this email." based on the communication data. The machine learning model M may specify each of the scheduled adjustment date and time, title, and memo from the text indicated by the communication data based on the instruction. The machine learning model M outputs scheduled adjustment data indicating what can be specified among the scheduled adjustment date and time, title, and memo.

[0068] [Scheduled Registration Data Acquisition Unit] The scheduled registration data acquisition unit 204 acquires scheduled registration data indicating the scheduled registrations registered in the scheduling management tool. In the present embodiment, since the scheduled registration data is stored in the scheduling management database DB3 stored in the scheduling management server 30, the scheduled registration data acquisition unit 204 acquires, from the scheduling management database DB3 stored in the scheduling management server 30, the scheduled registration data indicating the scheduled registrations in which the user logged in to the groupware is registered as a participant.

[0069] Note that the registration-scheduled data may be stored in a database other than the schedule management database DB3. The registration-scheduled data may be stored in a computer other than the schedule management server 30. The registration-scheduled data acquisition unit 204 may acquire the registration-scheduled data from the other database or the other computer. For example, the registration-scheduled data acquisition unit 204 acquires registration-scheduled data indicating the date and time of the registration schedule. The registration-scheduled data acquisition unit 204 may acquire registration-scheduled data indicating other information other than the date and time of the registration schedule.

[0070] [Display control unit] The display control unit 205 causes various screens to be displayed on the user terminal 40. For example, the display control unit 205 causes the screen to be displayed on the user terminal 40 by transmitting the display data of the screen to be displayed to the user terminal 40. When the screen is displayed on the browser, the display data of the screen is in HTML format. The display data of the screen may be in any format. For example, when the screen is displayed by a program dedicated to groupware, the display data of the screen may be in a format supported by the program.

[0071] For example, the display control unit 205 causes the communication screen SC1 to be displayed on the user terminal 40 based on the communication data. In the upper example of FIG. 2, the display control unit 205 causes the communication screen SC1 showing the content of the email to be displayed on the user terminal 40 based on the communication data of the email selected by the user. When the user selects the button B10 and the adjustment-scheduled data acquisition unit 203 acquires the adjustment-scheduled data, as shown in the lower part of FIG. 2, the display control unit 205 causes the communication screen SC1 including the content of the communication and the adjustment schedule indicated by the adjustment-scheduled data to be displayed on the user terminal 40.

[0072] For example, when the adjustment schedule data indicates the date and time of the adjustment schedule, as shown in the lower part of FIG. 2, the display control unit 205 causes the user terminal 40 to display a communication screen SC1 including the content of the communication and a display area A11 indicating the date and time of the adjustment schedule indicated by the adjustment schedule data. In the example shown in the lower part of FIG. 2, the content of the communication and the date and time of the adjustment schedule indicated by the adjustment schedule data are displayed on the same communication screen SC1. When the adjustment schedule data indicates other information other than the date and time of the adjustment schedule, the display control unit 205 causes the user terminal 40 to display a communication screen SC1 that also indicates the other information. For example, in the upper example of FIG. 3, the display control unit 205 causes the user terminal 40 to display a communication screen SC1 indicating the title acquired as other information.

[0073] For example, the display control unit 205 may cause the user terminal 40 to display a communication screen SC1 indicating the content of the communication based on the registration schedule data. For example, the display control unit 205 executes a determination as to whether the user's schedule is available based on the date and time of the adjustment schedule indicated by the adjustment schedule data and the date and time of the registration schedule indicated by the registration schedule data, and causes the user terminal 40 to display the communication screen SC1 based on the execution result of the determination. In the upper example of FIG. 3, the display control unit 205 displays the execution result of the above determination in the display area A11 based on the registration schedule data and the adjustment schedule data.

[0074] For example, the display control unit 205 determines whether the date and time of the adjustment schedule and the date and time of the registration schedule overlap. If they do not overlap, the display control unit 205 determines that the user's schedule is available. Note that the method by which the display control unit 205 determines whether the user's schedule is available is not limited to the above example. For example, the display control unit 205 may determine whether there is a registration schedule within a predetermined time (for example, 1 hour) before and after the date and time of the adjustment schedule. When there is no registration schedule within a predetermined time (for example, 1 hour) before and after the date and time of the adjustment schedule, the display control unit 205 determines that the user's schedule is available.

[0075] [Schedule Sending Unit] The scheduled transmission unit 206 transmits the scheduled adjustment data to a schedule management tool that cooperates with the communication tool. In the present embodiment, the scheduled transmission unit 206 transmitting the scheduled adjustment data to the schedule management server 30 corresponds to the scheduled transmission unit 206 transmitting the scheduled adjustment data to the schedule management tool. When the schedule management tool is managed by a computer other than the schedule management server 30, the scheduled transmission unit 206 may transmit the scheduled adjustment data to the other computer.

[0076] For example, the scheduled transmission unit 206 transmits the scheduled adjustment data to the schedule management tool based on the operations performed on the communication screen SC1. In the examples of FIGS. 2 and 3, the selection of the buttons B110 and B111 corresponds to the operations. The scheduled transmission unit 206 transmits the scheduled adjustment data to the schedule management tool when the user selects the button B111 after selecting the button B110 and checking the availability status. The operations that the user can perform on the communication screen SC1 may be arbitrary operations. The user's operations are not limited to the selection of the buttons B110 and B111. For example, the user may perform an operation on the date and time input form displayed in the display area A11, or may perform an operation on at least one of the input forms for the title and memo of the scheduled adjustment displayed in the display area A11.

[0077] In the present embodiment, when the scheduled adjustment displayed on the communication screen SC1 is not changed by the user, the scheduled transmission unit 206 transmits the scheduled adjustment data acquired by the scheduled adjustment data acquisition unit 203 to the schedule management tool. In the examples of FIGS. 2 and 3, when the user does not perform an operation to change the date and time on the date and time input form displayed in the display area A11, the scheduled transmission unit 206 transmits the scheduled adjustment data indicating the date and time to the schedule management tool as it is.

[0078] For example, when the adjustment schedule displayed on the communication screen SC1 is changed by the user, the scheduled transmission unit 206 transmits adjustment schedule data indicating the adjusted schedule changed by the user to the schedule management tool. In the examples of FIGS. 2 and 3, when the user performs an operation to change the date and time on the date and time input form displayed in the display area A11, the scheduled transmission unit 206 transmits the adjustment schedule data indicating the date and time after the change to the schedule management tool.

[0079] [Reply content generation unit] When the adjustment schedule data is transmitted to the schedule management tool, the reply content generation unit 207 generates reply content data indicating the reply content in the communication. For example, the reply content generation unit 207 generates reply content data indicating the date and time of the adjustment schedule registered as the registered schedule in the schedule management tool. The reply content generation unit 207 may also generate reply content data indicating other contents that were not registered in the schedule management tool.

[0080] In the lower example of FIG. 3, the reply content generation unit 207 generates reply content data indicating the reply content including the greeting registered as a template, the name and organization name of the logged-in user. The display control unit 205 causes the user terminal 40 to display the reply content on the communication screen SC1 by transmitting the reply content data generated by the reply content generation unit 207. It is assumed that data such as the user's affiliation and name in the reply content is stored in advance in the data storage unit 200. The name of the person to be adjusted in the reply content may be obtained from the email that is the source of the reply.

[0081] Note that the reply content generation unit 207 may generate reply content data based on the machine learning model M. In this case, the machine learning model M can execute the task of generating the reply content. For example, it is assumed that the machine learning model M has learned training data including the scheduled date and time for training and the correct reply content. The reply content generation unit 207 inputs data such as the scheduled date and time registered in the schedule management tool into the machine learning model M. The machine learning model M calculates the embedded representation of the data based on the parameters adjusted by learning, and outputs reply content data corresponding to the embedded representation. The reply content generation unit 207 acquires the reply content data from the machine learning model M.

[0082] For example, in the case of the dialogue-type machine learning model M, the reply content generation unit 207 may cause the machine learning model M to correct the reply content based on an instruction input by the user. For example, when the user inputs an instruction such as "Please use a more polite expression.", the reply content generation unit 207 causes the machine learning model M to correct the reply content so that the expression of the generated reply content becomes polite. In this case, the display control unit 205 transmits data indicating the corrected reply content to the user terminal 40 and displays the corrected reply content on the communication screen SC1. The user may repeat the correction of the reply content while repeating the dialogue with the machine learning model M. The user may also correct the reply content generated or corrected by the machine learning model M by himself / herself. Note that the user can input any instruction into the machine learning model M. The input instruction by the user is not limited to the above example. For example, the user may input instructions such as simplification of the reply text, language of the reply text, number of characters in the reply text, or other instructions. The machine learning model M may change the reply content based on these instructions.

[0083] [3-3. Functions Implemented by the Schedule Management Server] For example, the schedule management server 30 includes a data storage unit 300 and a schedule registration unit 301. The data storage unit 300 is realized by the storage unit 32. The schedule registration unit 301 is realized by the control unit 31.

[0084] [Data storage unit] The data storage unit 300 stores data necessary for schedule management. For example, the data storage unit 300 stores a schedule management database DB3.

[0085] FIG. 8 is a diagram showing an example of the schedule management database DB3. The schedule management database DB3 is a database in which registration schedule data indicating the registration schedules registered in the schedule management tool is stored. For example, the schedule management database DB3 stores a schedule ID and registration schedule data. In the present embodiment, storing the schedule ID and the registration schedule data in the schedule management database DB3 corresponds to registering a registration schedule in the schedule management tool.

[0086] The schedule ID is an ID for the schedule adjustment system 1 to uniquely identify a registration schedule. When a new registration schedule is registered in the schedule management tool, the communication server 20 generates a schedule ID for the new registration schedule. The communication server 20 generates registration schedule data indicating the new registration schedule. The communication server 20 registers a new registration schedule in the schedule management tool by storing the generated schedule ID and the registration schedule data in the schedule management database DB3. The registration schedule data is data indicating the specific content of the registration schedule. For example, the registration schedule data indicates the date and time, title, memo, participants, or a combination thereof of the registration schedule.

[0087] Note that the data stored in the data storage unit 300 is not limited to the above example. The data storage unit 300 can store any data. For example, the data storage unit 300 may store data of each screen of the schedule management tool.

[0088] [Schedule registration unit] The scheduled registration unit 301 registers the adjustment schedule in the schedule management tool based on the adjustment schedule data transmitted by the scheduled transmission unit 206. For example, the scheduled registration unit 301 generates registration schedule data for registering the adjustment schedule as a registration schedule based on the adjustment schedule data. If the user does not change the date and time of the adjustment schedule, the registration schedule data indicates the date and time indicated by the adjustment schedule data. The participants indicated by the registration schedule data are the logged-in users. The user may be able to specify participants from the communication screen SC1. The scheduled registration unit 301 issues a new schedule ID and stores the schedule ID and the registration schedule data in the schedule management database DB3 to register the adjustment schedule in the schedule management tool.

[0089] [Functions realized by the user terminal] For example, the user terminal 40 includes a data storage unit 400, a display control unit 401, and an operation reception unit 402. The data storage unit 400 is realized by the storage unit 42. Each of the display control unit 401 and the operation reception unit 402 is realized by the control unit 41.

[0090] [Data storage unit] The data storage unit 400 stores data necessary for schedule adjustment. For example, the data storage unit 400 stores a browser for displaying various screens of the schedule adjustment system 1. For example, the data storage unit 400 stores a program dedicated to groupware.

[0091] [Display control unit] The display control unit 401 causes the display unit 45 to display various screens in the schedule adjustment system 1. For example, the display control unit 401 causes the display unit 45 to display a screen such as the communication screen SC1 based on the display data received from the communication server 20.

[0092] [Operation reception unit] The operation reception unit 402 receives various operations in the schedule adjustment system 1. For example, it receives operations on the communication screen SC1 or the schedule management screen SC2. Data indicating the operation content received by the operation reception unit 402 is appropriately transmitted to the communication server 20.

[0093] [4. Processes Executed by the Schedule Adjustment System] FIGS. 9 and 10 are diagrams showing an example of the processes executed by the schedule adjustment system 1. The control units 21, 31, and 41 execute the programs stored in the storage units 22, 32, and 42, respectively, whereby the processes of FIGS. 9 and 10 are executed. It is assumed that the learning of the machine learning model M has been completed when the processes of FIGS. 9 and 10 are executed. The processes of FIGS. 9 and 10 are an example of the processes included in the schedule adjustment method of the present embodiment.

[0094] As shown in FIG. 9, when the user selects a specific email using the communication tool, the user terminal 40 executes a process for displaying the communication screen SC1 showing the details of the email selected by the user between the user terminal 40 and the communication server 20 (S1). In S1, the communication server 20 generates display data for the communication screen SC1 based on the communication database DB2 and transmits it to the user terminal 40. The user terminal 40 receives the display data for the communication screen SC1 and displays the communication screen SC1 on the display unit 45.

[0095] When the user selects button B10 on communication screen SC1, user terminal 40 sends a support request to communication server 20 for the user to request support by machine learning model M (S2). Communication server 20 receives the support request from user terminal 40 (S3). Communication server 20 refers to communication database DB2 and acquires the communication data of the email to be processed (displayed on communication screen SC1) (S4). Communication server 20 executes a process to set communication screen SC1 to the reply mode as shown in the lower part of FIG. 2 in communication with user terminal 40 based on the communication data (S5).

[0096] Communication server 20 determines whether the task of machine learning model M is schedule adjustment based on the communication data (S6). In S6, if it is determined that the task of machine learning model M is not schedule adjustment (S6:N), this process ends. In this case, the processes after S7 are not executed. Communication server 20 may display a message indicating that machine learning model M cannot support schedule adjustment on communication screen SC1. In this case, the user manually performs email reply and schedule registration to the schedule management tool from each of communication screen SC1 and schedule management screen SC2.

[0097] In S6, when it is determined that the task of the machine learning model M is schedule adjustment (S6: Y), the communication server 20 acquires adjustment schedule data indicating the date and time of the adjustment schedule based on the communication data and the machine learning model M (S7). In S7, the communication server 20 may also cause the machine learning model M to specify at least one of the schedule title and the memo. The communication server 20 executes a process for displaying the date and time of the adjustment schedule indicated by the adjustment schedule data in the display area A11 between the communication server 20 and the user terminal 40 (S8). In S7, when the adjustment schedule data is not acquired, a message indicating that the adjustment schedule data has not been acquired may be displayed in the display area A11 displayed in S8.

[0098] The user terminal 40 identifies the user operation performed from the operation unit 44 (S9). In S9, it is assumed that the date and time displayed in the display area A11 is changed or the button B110 is selected. When another operation is performed, a process corresponding to the other operation is executed and this process ends. In S9, when the user changes the date and time displayed in the display area A11 (S9: date and time change), the user terminal 40 changes the date and time of the adjustment schedule (S10). In S9, when the user selects the button B110 (S9: confirmation request), the user terminal 40 transmits a schedule confirmation request to the communication server 20 (S11). The confirmation request includes data indicating the date and time displayed in the display area A11.

[0099] The communication server 20 receives a request for schedule confirmation from the user terminal 40 (S12). The communication server 20 executes a process for confirming the date and time of the adjustment schedule with the schedule management server 30 (S13). In S13, the communication server 20 requests the schedule management server 30 for the registered schedule data of the user. The schedule management server 30 refers to the schedule management database DB3, acquires the registered schedule data of the user, and transmits it to the communication server 20. The communication server 20 determines whether the date and time of the adjustment schedule is available based on the date and time of the adjustment schedule and the date and time indicated by the registered schedule data of the user.

[0100] The communication server 20 determines whether the processing result of S13 indicates that the date and time of the adjustment schedule is available (S14). In S14, if it is determined that the date and time of the adjustment schedule is not available (S14:N), the process moves to FIG. 10, and the communication server 20 executes a process for displaying a message indicating that the date and time of the adjustment schedule is not available on the communication screen SC1 between the communication server 20 and the user terminal 40 (S15). In this case, the user changes the date and time of the adjustment schedule as necessary.

[0101] In S14, if it is determined that the date and time of the adjustment schedule is available (S14:Y), the process moves to FIG. 10, and the communication server 20 executes a process for displaying a message indicating that the date and time of the adjustment schedule is available on the communication screen SC1 based on the adjustment schedule data between the communication server 20 and the user terminal 40 (S16). In S7, if at least one of the schedule title and the memo is specified, at least one of the schedule title and the memo is displayed in the display area A11 by the processing of S16.

[0102] The user terminal 40 identifies the user operation performed from the operation unit 44 (S17). In S17, it is assumed that changes such as the date and time displayed in the display area A11, or the selection of the button B111 for transmitting the adjustment schedule, are made. If other operations are performed, the processing corresponding to the other operations is executed, and this processing ends. In S17, when the user changes the date and time, etc. displayed in the display area A11 (S17: change of date and time, etc.), the user terminal 40 changes the date and time, etc. of the adjustment schedule (S18). Then, the process proceeds to the process of S11, and the schedule is confirmed.

[0103] In S17, when the user selects the button B111 for transmitting the adjustment schedule (S17: transmission), the user terminal 40 transmits a transmission request for the adjustment schedule data to the communication server 20 (S19). The transmission request is data in a predetermined format for the user terminal 40 to request the communication server 20 to transmit the adjustment schedule data. The communication server 20 receives the transmission request from the user terminal 40 (S20). The communication server 20 transmits the adjustment schedule data to the schedule management server 30 (S21). In S21, if the user has not changed the date and time, etc. of the adjustment schedule, the communication server 20 transmits the adjustment schedule data acquired in S7 as it is. When the user has changed the date and time, etc. of the adjustment schedule, the communication server 20 transmits the adjustment schedule data indicating the changed date and time, etc.

[0104] The schedule management server 30 receives the adjustment schedule data from the communication server 20 (S22). The schedule management server 30 registers the adjustment schedule in the schedule management tool based on the adjustment schedule data (S23). In S23, the schedule management server 30 issues a new schedule ID and creates a new record in the schedule management database DB3. The schedule management server 30 generates registration schedule data for registering the adjustment schedule as a new registered schedule based on the adjustment schedule data, and stores the generated registration schedule data in the new record.

[0105] The communication server 20 generates reply content data based on the adjustment schedule data transmitted to the schedule management tool (S24). The communication server 20 executes a process for displaying the reply content indicated by the reply content data on the communication screen SC1 between the communication server 20 and the user terminal 40 (S25). The user terminal 40 identifies the user operation performed from the operation unit 44 (S26). In S26, the button B112 for changing the reply content or the button B12 for sending an email is selected. When another operation is performed, a process corresponding to the other operation is executed, and this process ends.

[0106] In S26, when the user selects the button B112 for changing the reply content (S26: change), the user terminal 40 executes a process for causing the machine learning model M to change the reply content between the user terminal 40 and the communication server 20 (S27), and returns to the process of S26. In S26, when the user selects the button B12 for sending an email (S26: send), the user terminal 40 executes a process for sending an email between the user terminal 40 and the communication server 20 (S28), and this process ends. The email may be sent by a known process.

[0107] [Summary of the Embodiment] The scheduling adjustment system 1 of the present embodiment acquires scheduling adjustment data indicating a scheduling adjustment based on communication data and a machine learning model M capable of language analysis. The scheduling adjustment system 1 transmits the scheduling adjustment data to a scheduling management tool that cooperates with a communication tool. Thereby, the scheduling adjustment system 1 can save the user the trouble of registering the scheduling adjustment in the scheduling management tool, so that the convenience of the user can be improved. For example, the person to be adjusted does not create an email under a predetermined format, but basically freely inputs text. Even if the scheduling adjustment system 1 receives an email without such a defined format, it can utilize language analysis by the machine learning model M to acquire the scheduling adjustment data. Since the scheduling adjustment system 1 can handle various emails, it can realize highly versatile scheduling adjustment support.

[0108] In addition, the scheduling adjustment system 1 registers the scheduling adjustment in the scheduling management tool based on the scheduling adjustment data transmitted by the scheduling transmission unit 206. Thereby, the scheduling adjustment system 1 can execute a series of processes from acquiring the scheduling adjustment data from the email to registering the scheduling adjustment in the scheduling management tool.

[0109] In addition, the scheduling adjustment system 1 acquires scheduling adjustment data indicating the date and time of the scheduling adjustment and other information of the scheduling adjustment. The scheduling adjustment system 1 can save the user the trouble of registering not only the date and time of the scheduling adjustment but also other information in the scheduling management tool, so that the convenience of the user can be effectively improved. For example, when the scheduling adjustment system 1 registers the title and memo of the schedule as other information of the scheduling adjustment in the scheduling management tool, the user does not need to manually input the title and memo of the schedule, so that the convenience of the user can be effectively improved. Even if the title and memo of the schedule specified by the machine learning model M are somewhat incorrect, the user only needs to make minor corrections to the errors, so that the scheduling adjustment system 1 can improve the convenience of the user.

[0110] In addition, when the adjustment schedule displayed on the communication screen SC1 is not changed by the user, the schedule adjustment system 1 transmits the adjustment schedule data acquired by the adjustment schedule data acquisition unit 203 to the schedule management tool. When the adjustment schedule displayed on the communication screen SC1 is changed by the user, the schedule adjustment system 1 transmits the adjustment schedule data indicating the adjusted schedule changed by the user to the schedule management tool. As a result, the user can check the adjustment schedule identified by the machine learning model M while checking the email on the communication screen SC1 and can change the adjustment schedule as needed. Therefore, the schedule adjustment system 1 can effectively improve the convenience of the user. For example, the user can register the adjustment schedule in the schedule management tool and reply to the email without going back and forth between the schedule management tool and the communication tool. For example, the user can also make a change such as securing a longer scheduled time in consideration of the content that cannot be read from the email text.

[0111] In addition, the schedule adjustment system 1 causes the user terminal 40 to display a communication screen SC1 indicating the content of the communication based on the scheduled registration data. The schedule adjustment system 1 registers the adjustment schedule in the schedule management tool based on the operation performed on the communication screen SC1. As a result, the schedule adjustment system 1 can register the adjustment schedule in the schedule management tool after allowing the user to confirm some information of the scheduled registration registered in the schedule management tool, making it easier to prevent double booking or being late, and effectively improving the convenience of the user.

[0112] In addition, the schedule adjustment system 1 executes a determination as to whether the user's schedule is available based on the date and time of the adjustment schedule indicated by the adjustment schedule data and the date and time of the scheduled registration indicated by the scheduled registration data, and causes the communication screen SC1 to be displayed on the user terminal 40 based on the execution result of the determination. As a result, the schedule adjustment system 1 can register the adjustment schedule in the schedule management tool after allowing the user to confirm some information of the scheduled registration registered in the schedule management tool, so that double booking or being late can be more reliably prevented.

[0113] Further, the schedule adjustment system 1 determines whether the task of the machine learning model M is schedule adjustment based on the communication data. When it is determined that the task is schedule adjustment, the schedule adjustment system 1 acquires the schedule adjustment data. Since the schedule adjustment system 1 can prevent unnecessary processing from being performed on emails that have nothing to do with schedule adjustment, the processing load on the communication server 20 is reduced. The schedule adjustment system 1 can also prevent a situation where text unrelated to schedule adjustment is obtained from an email unrelated to schedule adjustment and an incorrect schedule is registered in the schedule management tool.

[0114] Also, when the schedule adjustment data is sent to the schedule management tool, the schedule adjustment system 1 generates reply content data indicating the reply content in the communication. Thereby, the schedule adjustment system 1 can save the trouble of the user creating a reply to the adjustment partner, so that the convenience of the user can be effectively improved.

[0115] [6. Modification Example] Note that the present disclosure is not limited to the embodiments described above. It can be appropriately changed without departing from the spirit of the present disclosure.

[0116] [6-1. Modification Example 1] For example, the machine learning model M may identify a plurality of dates and times for schedule adjustment from an email. In this case, the display control unit 205 may select one of the plurality of dates and times and display the selected one date and time in the display area A11. However, in Modification Example 1, the display control unit 205 displays each of the plurality of dates and times in the display area A11. The user may select the date and time to be registered in the schedule management tool from among the plurality of dates and times displayed in the display area A11.

[0117] FIG. 11 is a diagram showing an example of the communication screen SC1 of Modification 1. The adjustment schedule data acquisition unit 203 of Modification 1 acquires adjustment schedule data indicating a plurality of dates and times of the adjustment schedule. For example, the machine learning model M identifies at least one date and time from the text indicated by the communication data based on the embedded representation calculated from the communication data input to itself. When the machine learning model M identifies a plurality of dates and times, the machine learning model M outputs adjustment schedule data indicating the plurality of dates and times. The adjustment schedule data acquisition unit 203 acquires the adjustment schedule data output from the machine learning model M.

[0118] The display control unit 205 of Modification 1 causes the user terminal 40 to display a communication screen SC1 including the content of the communication performed by the communication tool and the plurality of dates and times indicated by the adjustment schedule data. For example, the display control unit 205 causes the user terminal 40 to display a communication screen SC1 including a display area A11 in which any of the plurality of dates and times can be selected. In the example of the communication screen SC1 in FIG. 11, a case where a date and time is selected by a radio button is shown, but the display control unit 205 may display the date and time in a selectable manner by any method. For example, the display control unit 205 may use a checkbox, a pull-down menu, or other form to display the date and time in a selectable manner.

[0119] The schedule transmission unit 206 of Modification 1 transmits schedule adjustment data indicating the date and time selected by the user from among the plurality of dates and times indicated by the adjustment schedule data to the schedule management tool. For example, in the state of the communication screen SC1 in FIG. 11, when the user selects any of the plurality of dates and times and selects the button B110, the schedule transmission unit 206 checks the availability of the selected date and time and then transmits the schedule adjustment data indicating the date and time to the schedule management tool. Although it is different from the embodiment in that the adjustment schedule data indicating the date and time selected by the user among the plurality of dates and times is transmitted to the schedule management tool, the process for transmitting the adjustment schedule data is as described in the embodiment.

[0120] The scheduling adjustment system 1 of Modification Example 1 acquires scheduling adjustment data indicating a plurality of dates and times of the adjustment schedule. The scheduling adjustment system 1 displays a communication screen SC1 including the content of the communication performed by the communication tool and the plurality of dates and times indicated by the scheduling adjustment data. The scheduling adjustment system 1 transmits scheduling adjustment data indicating the date and time selected by the user from among the plurality of dates and times indicated by the scheduling adjustment data to the scheduling management tool. Even if it is an email including a plurality of dates and times, the scheduling adjustment system 1 can allow the user to select a desired date and time, so that the convenience of the user can be effectively enhanced.

[0121] [6-2. Modification Example 2] For example, the communication screen SC1 displayed based on the registration schedule data is not limited to the example of the embodiment. In Modification Example 2, another aspect of the communication screen SC1 displayed based on the registration schedule data will be described. The display control unit 205 of Modification Example 2 identifies the registration schedule at least on one of the front and the back of the date and time of the adjustment schedule indicated by the adjustment schedule data based on the registration schedule data, and causes the user terminal 40 to display the communication screen SC1 indicating the identified registration schedule.

[0122] FIG. 12 is a diagram showing an example of the communication screen SC1 of Modification Example 2. For example, the display control unit 205 determines whether there is a registration schedule within at least one of the front and the back of the date and time of the adjustment schedule (for example, 3 hours) based on the registration schedule data. When it is determined that there is a registration schedule within the predetermined time, the display control unit 205 causes the registration schedule to be displayed in the display area A11. In Modification Example 2, the case where the registration schedules on both the front and the back of the date and time of the adjustment schedule are the determination targets is taken as an example, but the registration schedule on either the front or the back of the date and time of the adjustment schedule may be the determination target.

[0123] In the example of FIG. 12, since the display control unit 205 identified a registration schedule that ends 30 minutes before the scheduled adjustment start date and time and a registration schedule that starts 1 hour after the scheduled adjustment date and time, these two registration schedules are displayed in the display area A11. The user checks the display area A11 and changes the scheduled adjustment date and time as necessary. The user may change the registration schedule date and time without changing the scheduled adjustment date and time. Note that the display control unit 205 may target registration schedules on the same date as the scheduled adjustment date and time for determination. The period targeted by the display control unit 205 is not limited to the example described in Modification 2 and may be any period.

[0124] Based on the registration schedule data, the schedule adjustment system 1 of Modification 2 identifies registration schedules at least on one of the front and back of the scheduled adjustment date and time indicated by the adjustment schedule data, and causes the communication screen SC1 showing the identified registration schedules to be displayed on the user terminal 40. Thereby, the user can register the adjustment schedule in the schedule management tool while checking the registration schedules at least on one of the front and back of the scheduled adjustment date and time, so the schedule adjustment system 1 can effectively improve the convenience for the user. For example, the user can determine the scheduled adjustment date and time considering the travel time from the location of the registration schedule before the adjustment schedule, or can determine the scheduled adjustment date and time considering the travel time to the location of the registration schedule after the adjustment schedule.

[0125] [6-3. Modification 3] For example, the display control unit 205 may cause the communication screen SC1 including mobility information indicating mobility to be displayed on the user terminal 40 based on the scheduled adjustment date and time and location indicated by the adjustment schedule data and the scheduled registration date and time and location indicated by the registration schedule data. The mobility information is information indicating whether the user can make it from the location of the scheduled adjustment to the location of the next registration schedule after the scheduled adjustment. The mobility information may indicate whether the user can make it from the registration schedule before the scheduled adjustment to the location of the scheduled adjustment.

[0126] FIG. 13 is a diagram showing an example of the communication screen SC1 of Modification 3. For example, the display control unit 205 calculates the travel time between the location scheduled for adjustment and at least one of the locations scheduled for registration before and after it. The calculation of the travel time can utilize a known route search algorithm. The route search algorithm may calculate the travel time when the user uses a means of transportation such as a train or a bus. The route search algorithm may be one adopted by a known map application. The display control unit 205 may calculate the travel time using a known map application.

[0127] For example, each of the location scheduled for adjustment indicated by the adjustment schedule data and the location scheduled for registration indicated by the registration schedule data may indicate an address or latitude and longitude, or the route search algorithm may indicate information (e.g., company name or building name) from which the address or latitude and longitude can be estimated. In Modification 3, the machine learning model M identifies the location scheduled for adjustment from the body of the email. The training data of the machine learning model M shows the training locations described in the body of the training email. The training data is learned by the machine learning model M. The machine learning model M analyzes the body of the email indicated by the communication data, identifies the location described in the body, and outputs adjustment schedule data.

[0128] For example, the display control unit 205 calculates the time interval between the date and time scheduled for adjustment and at least one of the dates and times scheduled for registration before and after the adjustment schedule. At least one of the registration schedules before and after the adjustment schedule may be specified in the same manner as in Modification 2. The time interval may be the interval between the start date and time of the adjustment schedule and the end date and time of the registration schedule before the adjustment schedule, or the interval between the end date and time of the adjustment schedule and the start date and time of the registration schedule after the adjustment schedule. The display control unit 205 executes a determination as to whether or not the calculated travel time is equal to or greater than the calculated time interval. The display control unit 205 causes the travel possibility information to be displayed in the display area A11 based on the execution result of the determination.

[0129] For example, when the calculated travel time is less than the calculated time interval, the display control unit 205 causes the display area A11 to display mobility information indicating that movement is possible. In the example of FIG. 13, the travel time to the next scheduled registration location to be adjusted is 15 minutes. The time interval between the adjustment schedule and the next registration schedule is 1 hour. Since the travel time of 15 minutes is shorter than the time interval of 1 hour, the display control unit 205 causes the display area A11 to display mobility information indicating that it is possible to move to the next scheduled registration location. Conversely, when the calculated travel time is greater than or equal to the calculated time interval, the display control unit 205 causes the display area A11 to display mobility information indicating that movement is not possible.

[0130] The schedule adjustment system 1 of Modification 3 causes the user terminal 40 to display a communication screen SC1 including mobility information indicating mobility based on the date and time and location of the adjustment schedule indicated by the adjustment schedule data and the date and time and location of the registration schedule indicated by the registration schedule data. Since the schedule adjustment system 1 can allow the user to grasp the mobility on the communication screen SC1, the convenience of the user can be improved. For example, the schedule adjustment system 1 can prevent the user from being late for the adjustment schedule or the next registration schedule.

[0131] [6-4. Modification 4] For example, an email with repeated replies may include a plurality of dates and times presented by the user in the past, or dates and times with inconsistent schedules. The email may also include other dates and times such as the reply date and time of the email. In this case, since the machine learning model M may identify incorrect dates and times, only the portion of the email with a high probability of being the date and time of the adjustment schedule may be the processing target of the machine learning model M.

[0132] The adjustment-scheduled data acquisition unit 203 of Modification Example 4 determines a processing target portion to be processed by the machine learning model M from the communication data, and acquires adjustment-scheduled data based on the machine learning model M and the processing target portion. The processing target portion is a part of the text indicated by the communication data. For example, the processing target portion is a relatively new part of the text indicated by the communication data. In the case of an email, the processing target portion is a relatively new part of the reply content described in the email. Similarly, for other communication tools such as chat, the processing target portion may be a relatively new part of the reply content.

[0133] In Modification Example 4, an example is given where the latest reply portion among emails with repeated replies corresponds to the processing target portion. In the example of the upper email in FIG. 2, the portion before the portion indicating the user's reply (the portion before the line starting with "2023 / 10 / 20 10:45" in FIG. 2) is the latest reply content, so this portion corresponds to the processing target portion. For example, the adjustment-scheduled data acquisition unit 203 specifies the processing target portion based on the position of the quotation mark indicating the reply (the symbol "|" in FIG. 2).

[0134] Note that the adjustment-scheduled data acquisition unit 203 may specify the portion up to a predetermined number of lines from the top as the processing target portion instead of the position of the quotation mark, or may specify the portion up to the line immediately before the line where the email address is found in the body of the email as the processing target portion. The adjustment-scheduled data acquisition unit 203 may use, as the processing target portion, other portions of the email excluding portions such as the reply date and time.

[0135] For example, the adjustment schedule data acquisition unit 203 inputs the processing target part among the texts indicated by the communication data into the machine learning model M. The machine learning model M divides the processing target part into a plurality of tokens and calculates an embedding representation. Based on the calculated embedding representation, the machine learning model M outputs adjustment schedule data indicating the date and time of the adjustment schedule included in the processing target part. The adjustment schedule data acquisition unit 203 acquires the adjustment schedule data output from the machine learning model M. Although it is different from the embodiment in that the entire email is not subject to the processing of the machine learning model M, the internal processing of the machine learning model M may be the same as that of the embodiment.

[0136] In addition, when the machine learning model M cannot identify the date and time of the adjustment schedule from the processing target part, the adjustment schedule data acquisition unit 203 may input all or part of other parts other than the processing target part among the texts indicated by the communication data into the machine learning model M. For example, the adjustment schedule data acquisition unit 203 may input the part indicating the most recent reply of the email (in FIG. 2, the part where the user Ayahou Taro replied at "2023 / 10 / 20 10:45") into the machine learning model M. That is, the adjustment schedule data acquisition unit 203 may successively input relatively new processing target parts into the machine learning model M until the machine learning model M identifies some date and time.

[0137] The schedule adjustment system 1 of Modification 4 determines the processing target part to be processed by the machine learning model M from the communication data, and acquires the adjustment schedule data based on the machine learning model M and the processing target part. The schedule adjustment system 1 can accurately acquire the adjustment schedule data. For example, the schedule adjustment system 1 can prevent an incorrect date and time from being identified from an email in which replies are repeated.

[0138] [6-5. Modification 5] For example, in the embodiment, the case where adjustment schedule data is acquired from the emails managed by the groupware is taken as an example. The user may perform schedule adjustment not only using emails but also using other communication tools such as chats or threads. Therefore, the schedule adjustment system 1 may be configured to be able to handle each of a plurality of communication tools.

[0139] The communication data acquisition unit 201 of Modification 5 acquires the communication data of each of a plurality of communication tools. The plurality of communication tools may be tools of the same groupware, or may be tools that have no particular relation to the groupware. In Modification 5, the case where the communication data acquisition unit 201 acquires the communication data of a first communication tool which is an email and a second communication tool which is a chat is taken as an example. The communication data acquisition unit 201 may acquire communication data from each of three or more communication tools. The communication data of each communication tool may be stored in the data storage unit 200, or may be stored in a computer other than the communication server 20.

[0140] The adjustment schedule data acquisition unit 203 of Modification Example 5 identifies the communication tool from which communication data has been acquired among a plurality of communication tools, and acquires adjustment schedule data based on the identified communication tool. For example, when a separate machine learning model M is prepared for each communication tool, the adjustment schedule data acquisition unit 203 acquires adjustment schedule data based on the machine learning model M corresponding to the communication tool from which communication data has been acquired. It is assumed that training data created from the communication data of the communication tool corresponding to the machine learning model M has been learned by each machine learning model M. Therefore, the tendency of the communication tool corresponding to each machine learning model M has been learned by each machine learning model M. The internal processing of each machine learning model M may be the same as the internal processing described in the embodiment.

[0141] For example, when one machine learning model M can correspond to a plurality of communication tools, the adjustment schedule data acquisition unit 203 inputs not only the communication data but also data indicating the communication tool from which the communication data has been acquired to the machine learning model M. The input part of the training data of the machine learning model M includes data indicating the communication tool. It is assumed that the machine learning model M is learned so that when training communication data and data indicating the communication tool corresponding to the communication data are input, the date and time etc. included in the text indicated by the communication data are output. The adjustment schedule data acquisition unit 203 acquires the adjustment schedule data output from the machine learning model M.

[0142] The scheduling adjustment system 1 of Modification Example 5 identifies the communication tool from which communication data has been acquired among a plurality of communication tools, and acquires adjustment schedule data based on the identified communication tool. As a result, since the scheduling adjustment system 1 can correspond to a plurality of communication tools, the convenience for the user can be effectively enhanced. For example, when an email has a formal text and a chat has a casual text, the machine learning model M can understand the difference between these texts and identify the adjustment schedule, so the accuracy of the machine learning model M in identifying the adjustment schedule is improved.

[0143] [6-6. Other Modification Examples] For example, two or more of Modification Examples 1 to 5 may be combined.

[0144] For example, although the case where the scheduling adjustment system 1 includes the learning terminal 10 has been cited as an example, the scheduling adjustment system 1 may perform learning of the machine learning model M without including the learning terminal 10. In this case, the scheduling adjustment system 1 may acquire adjustment schedule data based on a known machine learning model M provided free of charge or for a fee. The scheduling adjustment system 1 may transmit the adjustment schedule data to the schedule management tool without displaying the adjustment schedule indicated by the adjustment schedule data on the communication screen SC1. The scheduling adjustment system 1 may transmit the adjustment schedule data to the schedule management tool without checking whether the user is available.

[0145] For example, at least one of the communication tool and the schedule management tool may be a tool not related to groupware. In the embodiment and Modification Examples 1 to 5, the case where the user presents the date and time that are candidates for the adjustment schedule to the adjustment partner has been cited as an example. However, even when the adjustment partner presents the date and time that are candidates for the adjustment schedule to the user, the scheduling adjustment system 1 may acquire adjustment schedule data indicating the date and time of the adjustment schedule and the like by the same processing as in the embodiment and Modification Examples 1 to 5.

[0146] For example, the operator providing the communication tool to the user and the operator providing the schedule management tool to the user may be different. In this case, these operators are assumed to cooperate with each other. The communication server 20 managed by a certain operator transmits the adjustment schedule data to the schedule management server 30 managed by another operator. The schedule management server 30 managed by another operator registers the adjustment schedule in the schedule management tool managed by the said other operator based on the adjustment schedule data. Also in such a case, by executing the same processing as in the embodiments and modification examples 1 to 5, a series of processes of acquisition of the adjustment schedule data, transmission of the adjustment schedule data, and registration of the adjustment schedule may be executed. In this case, the schedule adjustment system 1 may not include the schedule registration unit 301. The schedule registration unit 301 may be realized by a system external to the schedule adjustment system 1.

[0147] For example, one server computer may manage both the communication tool and the schedule management tool. In this case, within the said server computer, the adjustment schedule data is transmitted from the communication tool to the schedule management tool. This transmission is performed by inter-process communication or the like. A certain server computer transmits the adjustment schedule data to the schedule management tool stored in itself from the communication tool stored in itself by internal communication. When the adjustment schedule data is transmitted by internal communication, this server computer registers the adjustment schedule in the schedule management tool.

[0148] For example, the functions described as being realized by the communication server 20 may be realized by the user terminal 40. In this case, the said functions may be realized by a browser script or an application installed in the user terminal 40. For example, each function may be shared by a plurality of computers or may be realized by one computer.

Explanation of Reference Numerals

[0149] 1 Scheduling adjustment system, 10 Learning terminals, 11, 21, 31, 41 Control units, 12, 22, 32, 42 Memory units, 13, 23, 33, 43 Communication units, 20 Communication server, 30 Scheduling management server, 40 User terminal, 14, 44 Operation units, 15, 45 Display units, M Machine learning model, N Network, A11 Display area, B10, B12, B110, B111, B112 Buttons, DB1 Training database, DB2 Communication database, DB3 Scheduling management database, SC1 Communication screen, SC2 Scheduling management screen, 100 Data storage unit, 101 Learning unit, 200 Data storage unit, 201 Communication data acquisition unit, 202 Task determination unit, 203 Scheduled adjustment data acquisition unit, 204 Scheduled registration data acquisition unit, 205 Display control unit, 206 Scheduling transmission unit, 207 Reply content generation unit, 300 Data storage unit, 301 Scheduled registration unit, 400 Data storage unit, 401 Display control unit, 402 Operation reception unit.

Claims

[Claim 1] a communication data acquisition unit that acquires communication data indicating the content of communication performed using the communication tool; an adjustment schedule data acquisition unit that acquires adjustment schedule data indicating an adjustment schedule to be adjusted by the communication based on the communication data and a machine learning model capable of language analysis; a schedule transmission unit that transmits the adjusted schedule data to a schedule management tool that cooperates with the communication tool; A scheduling system including:

Images