WO2023095215A1 - Complaint occurrence prediction system, complaint occurrence prediction method, and program - Google Patents

Abstract

[Problem] To make it possible to improve the handling of dissatisfaction and complaints from trading partners, and to improve sales and repeat rate. [Solution] A complaint occurrence prediction system that predicts the occurrence of complaints acquires a plurality of first emails received from a trading partner, determines a first complaint risk index, indicating the risk of complaints received from the trading partner, for each of the plurality of acquired first emails, trains by associating each of the plurality of acquired first emails with the first complaint risk index determined for same, generates a trained model on the basis of a training result, acquires a second email newly received from the trading partner, predicts a second complaint risk index for the acquired second email on basis of the generated trained model, the second complaint risk index indicating the risk of complaints to be received from the trading partner in the future; and issues an alert, in which the acquired second email is associated with the second complaint risk index, when the predicted second complaint risk index satisfies a first condition.

Description

Complaint Occurrence Prediction System, Complaint Occurrence Prediction Method and Program

The present invention relates to technology effective in predicting the occurrence of claims.

In recent years, technology related to claim prediction has attracted attention.
For example, in Patent Document 1, in order to make it possible to predict the occurrence of complaints related to the living environment, past measured values representing the state of the living environment and the presence or absence of complaints when the past measured values were obtained are used. technology is provided to learn
Yet another technique is provided in US Pat. No. 6,200,302 for estimating policy loss reserves and confidence intervals using policy and claim level detailed predictive modeling.

JP 2020-186850 A Japanese Patent Publication No. 2008-512798

2. Description of the Related Art It is a daily practice to receive e-mails with various contents from business partners (for example, companies, general consumers, and organizations). Depending on the content of the e-mail, misunderstandings by employees and staff can lead to complaints. It is important to have a common understanding of the degree of
Therefore, there is a demand for a technique for predicting a complaint risk index, which is an index of the risk of complaints received from business partners in the future, and notifying employees and staff of the target e-mail.
However, with the techniques described in Patent Documents 1 and 2, it was not possible to predict the risk index of complaints received from business partners in the future with respect to target emails and notify employees and staff.
Therefore, the present invention focuses on a mechanism for predicting the risk index of complaints received from business partners in the future for target e-mails and notifying employees and staff.

The inventor of the present invention predicts the risk index of complaints received from business partners in the future for target e-mails and notifies employees and staff, thereby improving the handling of complaints and complaints from business partners and increasing sales. An object of the present invention is to provide a complaint occurrence prediction system, a complaint occurrence prediction method, and a program that enable an improvement in the repeat rate.

The present invention is a claim occurrence prediction system that predicts the occurrence of a claim,
a first acquisition unit that acquires a plurality of first emails received from business partners;
a determination unit that determines a first complaint risk index indicating a risk of complaints received from the business partner for each of the plurality of first emails obtained;
a learning unit that learns by associating each of the plurality of first emails obtained with the first complaint risk index determined for each;
a generation unit that generates a trained model based on the learning result;
a second acquisition unit that acquires a second mail newly received from the trading partner;
a prediction unit that predicts, based on the generated learned model, a second complaint risk index indicating the risk of complaints received from the business partner in the future for the acquired second e-mail;
a notification unit that notifies an alert that associates the obtained second email with the second complaint risk index when the predicted second complaint risk index satisfies a first condition;
To provide a claim occurrence prediction system comprising

According to the present invention, a complaint occurrence prediction system for predicting the occurrence of a complaint acquires a plurality of first emails received from a business partner, and for each of the acquired plurality of first emails, receives from the business partner determining a first complaint risk index indicating the risk of a complaint to be received, learning by associating each of the plurality of first emails obtained with the first complaint risk index determined for each, and based on the learning result generating a learned model, obtaining a second mail newly received from the business partner, and receiving the obtained second mail from the business partner in the future based on the generated learned model; A second complaint risk index indicating the degree of complaint risk is predicted, and when the predicted second complaint risk index satisfies a first condition, the obtained second e-mail is associated with the second claim risk index. Notification of alerts.

Although the present invention is in the category of systems, methods and programs have similar actions and effects.

According to the present invention, it is possible to improve customer dissatisfaction and complaint handling, and improve sales and repeat rate.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure explaining the outline|summary of the complaint occurrence prediction system 1. FIG. 1 is a diagram showing a functional configuration of a complaint occurrence prediction system 1; FIG. 4 is a diagram showing a flowchart of learning processing executed by the complaint occurrence prediction system 1. FIG. FIG. 10 is a diagram showing a flowchart of notification destination setting processing executed by the complaint occurrence prediction system 1; FIG. 10 is a diagram showing a flowchart of notification destination change processing executed by the complaint occurrence prediction system 1; 4 is a diagram showing a flowchart of alert notification processing executed by the complaint occurrence prediction system 1. FIG. 4 is a diagram schematically showing an example of a first alert 40; FIG. It is a figure which shows an example of the 2nd alert 50 typically. 4 is a diagram showing a flowchart of mail output processing executed by the complaint occurrence prediction system 1. FIG. 6 is a diagram schematically showing an example of a management screen 60; FIG. 7 is a diagram schematically showing an example of a mail screen 70; FIG. FIG. 10 is a diagram showing a flowchart of rearrangement execution processing executed by the complaint occurrence prediction system 1. FIG. It is the figure which showed an example of the rearrangement result typically. It is the figure which showed an example of the rearrangement result typically. 4 is a diagram showing a flowchart of re-learning processing executed by the complaint occurrence prediction system 1. FIG.

Hereinafter, with reference to the attached drawings, the modes for carrying out the present invention (hereinafter referred to as embodiments) will be described in detail. In subsequent figures, the same numbers or symbols are attached to the same elements throughout the description of the embodiments.

[Basic Concept/Basic Configuration]
FIG. 1 is a diagram for explaining an outline of a complaint occurrence prediction system 1. As shown in FIG. The complaint occurrence prediction system 1 is a system that predicts the occurrence of complaints and includes at least a computer 10 .
In this embodiment, the complaint occurrence prediction system 1 is connected to a computer 10, a mail server 20 that manages the transmission and reception of mail between a user and a business partner, and a user terminal 30 that is managed by the user so that data communication is possible. It is a system that

In this embodiment, it is assumed that the mail server 20 is connected to the computer 10 and the user terminal 30 via a POP (Post Office Protocol) connection so that data communication is possible. 10, to transmit the sent and received mails.

An overview of the processing steps when the complaint occurrence prediction system 1 notifies an alert that associates an email with the degree of complaint risk will be described based on FIG.

The computer 10 acquires a plurality of first mails received from business partners (step S1).
The computer 10 acquires, from the mail server 20, a plurality of mails that have been received by one or more users from business partners as first mails.

The computer 10 determines a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of acquired first mails (step S2).
The first claim risk index is, for example, a numerical value from 1 to 10,000.
The computer 10 accepts input of the numerical value of the complaint risk level for each of the first mails from the user or the system administrator, and determines the received numerical value as the first complaint risk index. In addition, the computer 10 determines the first complaint risk index according to the contents of complaints that have occurred or are about to occur in each of the first mails in the past. Also, the computer 10 determines the first complaint risk index according to the mail contents of the first mail.

The computer 10 learns by associating each of the plurality of acquired first emails with the first complaint risk index determined for each (step S3), and based on the learning result, generates a trained model (step S4 ).
The computer 10, for example, uses the determined first complaint risk index as training data, and executes supervised learning in which the content of each of the obtained first emails is associated with the determined first complaint risk index, Generate a trained model based on the training results.

The computer 10 acquires the second mail newly received from the trading partner (step S5).
When the mail server 20 receives a new mail from a business partner, the computer 10 acquires this mail from the mail server 20 as a second mail.

Based on the generated learned model, the computer 10 predicts a second complaint risk index indicating the risk of future complaints from business partners for the obtained second mail (step S6).
The computer 10 refers to the generated learned model and predicts the second complaint risk index of the second mail based on the obtained mail contents of the second mail.

If the predicted second complaint risk index satisfies the first condition, the computer 10 notifies an alert that associates the obtained second mail with the second complaint risk index (step S7).
The first condition is, for example, that the second claim risk index is 500 or more.
If the predicted second complaint risk index satisfies the first condition, the computer 10 creates an alert that associates the obtained second e-mail with the second complaint risk index, and sends the created alert to a preset value. The user terminal 30 managed by the user of the notification destination is notified.

According to such a complaint occurrence prediction system 1, it is possible to improve customer dissatisfaction and complaint handling, and improve sales and repeat rate.

[Function configuration]
Based on FIG. 2, the functional configuration of the complaint occurrence prediction system 1 will be described.
The complaint occurrence prediction system 1 includes at least a computer 10, and the computer 10 includes a mail server 20 for managing transmission and reception of mail between a user and a business partner, a user terminal 30 managed by the user, a public line network, Via a network 9 such as an intranet, they are connected so as to be capable of data communication.
In addition to the computer 10, the complaint prediction system 1 may include a mail server 20, a user terminal 30, other terminals and devices. In this case, the complaint prediction system 1 executes each process described later by any one or a combination of included computers, terminals, devices, and the like.

The computer 10 is a computer, a personal computer, or the like having a server function for predicting the occurrence of complaints.
The computer 10 may be realized by, for example, one computer, or may be realized by a plurality of computers like a cloud computer. A cloud computer in this specification is a computer that uses any computer in a scalable manner to perform a specific function, or includes multiple functional modules to realize a certain system, and uses the functions in a free combination. It can be anything.

The computer 10 includes a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), etc. as a control unit, and a communication unit to communicate with other terminals, devices, etc. It includes a device for enabling communication, a first acquisition unit 11 that acquires a first mail, a second acquisition unit 12 that acquires a second mail, a notification unit 13 that notifies an alert, and the like.
The computer 10 also includes a data storage unit such as a hard disk, a semiconductor memory, a storage medium, or a memory card as a storage unit.
Further, the computer 10 includes, as processing units, various devices for executing various processes, a determination unit 14 for determining the first complaint risk index, a learning unit 15 for learning by associating the first mail with the first complaint risk index, a learning It comprises a generation unit 16 that generates a learned model based on the result, a prediction unit 17 that predicts the second claim risk index, and the like.

In the computer 10, by reading a predetermined program by the control unit, in cooperation with the communication unit, a first acquisition module, a notification destination reception module, a change reception module, a second acquisition module, a notification module, a management screen output module, A selection reception module, a mail output module, a rearrangement reception module, a rearrangement result output module, and a correction reception module are implemented.
Also, in the computer 10, the control unit reads a predetermined program to realize a learned model storage module and a notification destination storage module in cooperation with the storage unit.
Further, in the computer 10, the control unit reads a predetermined program, and cooperates with the processing unit to perform an exclusion module, a determination module, a learning module, a generation module, a setting module, a change module, an unsolicited e-mail determination module, and a prediction module. , a first condition determination module, a second condition determination module, an acceptance determination module, and an execution module.

The mail server 20 may be server software or a server computer for delivering general e-mails for delivering mails between users and business partners, and detailed description thereof will be omitted.

The user terminal 30 is a terminal device such as a mobile terminal such as a mobile phone, a smartphone, a tablet terminal, or a personal computer managed by the user, and includes a CPU, GPU, RAM, ROM, etc. as a terminal control unit, A device or the like for enabling data communication with the computer 10 and the mail server 20 is provided, and as an input/output unit, various devices and the like for displaying alerts, management screens and emails, and inputting/outputting data are provided.

Each process executed by the complaint occurrence prediction system 1 will be described below together with the process executed by each module described above.

[Learning processing executed by computer 10]
Learning processing executed by the computer 10 will be described with reference to FIG. This figure is a diagram showing a flowchart of the learning process executed by the computer 10 . This learning process includes the acquisition process of the first e-mail (step S1), the process of determining the first complaint risk index (step S2), and the learning process of learning by associating the first e-mail with the first complaint risk index (step S2). S3), details of the process of generating a learned model based on the learning result (step S4).

The first acquisition module acquires a plurality of first emails received from business partners (step S10).
A first acquisition module acquires a plurality of emails that have been received by one or more users from business partners as first emails.
The mail server 20 transmits to the computer 10 a plurality of first mails that the user has received so far from business partners based on a request from the first acquisition module.
The first acquisition module acquires the first mail by receiving this first mail.
It should be noted that the first acquisition module can also be configured to acquire, in addition to the first email, the email sent by the user to the business partner.
Also, the number of first emails acquired by the first acquisition module is not particularly limited, but may be any number necessary for use in the learning process described later. Also, the first mail acquired by the first acquisition module may satisfy predetermined conditions such as a predetermined period and a predetermined business partner.

The exclusion module excludes unwanted e-mails from the plurality of obtained first e-mails (step S11).
Spam e-mails are advertising e-mails, fictitious billing e-mails, fraudulent billing e-mails, virus e-mails, money-making e-mails, chain e-mails, spoofing e-mails, phishing e-mails, etc., which are sent from third parties other than business partners, regardless of the intention of the user. It was sent randomly.
The exclusion module excludes unwanted e-mails from the plurality of acquired first e-mails by executing processing related to filtering existing unwanted e-mails.

The determination module determines a first complaint risk index indicating the risk of complaints received from the business partner for each of the obtained first mails (step S12).
The first claim risk index is, for example, a numerical value from 1 to 10,000.
The determination module determines this first claim risk index for each spam-filtered first email.
The determination module receives input of a complaint risk level numerical value for each first mail from a user or system administrator, and determines the received numerical value as a first complaint risk index for each first mail. In this case, the determination module accepts the input of this numerical value via the user terminal 30, physical device or virtual device connected to the computer 10, and converts the received numerical value to the first claim of each of the first mails. Decide on a risk index.
In addition, the determination module determines the risk of the first complaint for each of the first emails according to the content of complaints that have occurred in each of the first emails in the past, and complaints that were likely to occur but did not actually result in complaints. Determine the exponent. In this case, the determination module determines the degree of risk of complaints that have occurred or are about to occur with respect to the obtained first mail as the first complaint risk index. The degree of risk of complaints is the impact of complaints such as deterioration of business performance, deterioration of corporate image, increased risk of unreasonable demands, prolonged response, increased loss, etc., and is a quantification of this impact. The determination module determines this numerical value of risk in each of the pre-stored first mails as a first claim risk index for each of the first mails.
Also, the determining module determines a first complaint risk index for each of the first emails according to email contents such as subject, body, sender's email address, receiver's email address, etc. in each of the first emails. The mail content is obtained as a result of analysis of the first mail by character recognition or the like. In this case, the determination module refers to a registered database in which the contents of the mail, complaints that have arisen or are about to occur from the contents of this mail, and numerical values of the risks are associated with each other and registered. 1 Determine the claim risk index. The determination module identifies the maximum value, average value, sum, etc. of the risk values in the database as the risk values in the first mail, and uses the identified risk values as the first claim risk index. Determined as
It should be noted that the method by which the determination module determines the first claim risk index for each of the first mails is not limited to the example described above.

The learning module learns by associating each of the acquired first mails with the first complaint risk index determined for each (step S13).
Learning methods include, for example, machine learning such as supervised learning, unsupervised learning, and reinforcement learning, and deep learning such as convolutional neural networks, recurrent neural networks, and long/short-term memory. In this embodiment, the learning module performs machine learning by supervised learning as a learning method.
The learning module uses the determined first complaint risk index as training data, and performs supervised learning in which the content of each of the obtained first emails is associated with the determined first complaint risk index.
Note that the learning method executed by the learning module is not limited to the example described above.

The generation module generates a learned model based on the learning result (step S14).
The generation module generates, for example, a regression model or a classification model based on learning results as a trained model. The method by which the generation module generates the learned model based on the learning result may be any existing method, and detailed description thereof will be omitted.

The learned model storage module stores the generated learned model (step S15).

The above is the learning process.
The computer 10 uses the learned model generated by the learning process described above to execute the process described later.
Note that the computer 10 can be configured to execute the processing described later without storing the trained model in the learning processing. In this case, the generated trained model can be used as it is in the processing described later.

[Notification Destination Setting Process Executed by Computer 10]
Based on FIG. 4, the notification destination setting process executed by the computer 10 will be described. This figure is a diagram showing a flowchart of the notification destination setting process executed by the computer 10 . This notification destination setting process is a process related to the above-described alert notification process (step S7), and can be executed before or after the above-described learning process.

The notification destination receiving module receives an input of an alert notification destination (step S20).
The notification destination is, for example, the user to whom the obtained email is sent, the user to whom the email is sent, the user related to the user to whom the email is sent, who belong to the same team or department, etc., and the email of the obtained email. These are users related to content and users who provide services to business partners such as customer support departments.
The notification destination receiving module outputs to the user terminal 30 a predetermined input screen for receiving the input of the notification destination. The user terminal 30 receives this input screen and displays it on its own display unit. The notification destination reception module receives input of a notification destination through a predetermined input screen displayed on the user terminal 30 .
The user terminal 30 accepts the input of the user's email address to be notified by accepting the direct input of the user's email address. This may accept direct entry of one e-mail address, or may accept direct entry of a plurality of e-mail addresses. For example, direct input of the e-mail address of the user to whom the client sends e-mail, or the e-mail address of another user such as the user's boss or the person in charge of the department to which the user belongs is accepted. Alternatively, direct entry of part or all of the mail addresses of this user and other users is accepted.
In addition, the notification destination reception module creates groups in which users are grouped in advance for each predetermined classification (for example, the same team, the same department, the complaint handling team, the complaint handling department, the person in charge, the boss, and all users). Register the email address of each user who belongs to this group to the group you created. The notification destination receiving module outputs the created group to the user terminal 30 together with a predetermined input screen.
The user terminal 30 accepts the input of the mail address of the user to be notified by accepting the selection input of this group. This may accept a selection input for one group, or may accept a selection input for a plurality of groups.
The user terminal 30 transmits the received notification destination to the computer 10 . In the case of an e-mail address, it is the received e-mail address. In the case of a group, it is the name of the group, an identifier such as a management number, and the e-mail address registered in the received group.
The user terminal 30 transmits these received notification destinations to the computer 10 .
By receiving these notification destinations, the notification destination reception module receives the input of alert notification destinations.

In step S20, the notification destination receiving module may be configured to receive an input of an alert notification destination with other conditions added. Specifically, the user terminal 30 can be configured to receive an input of a notification destination according to the content of the mail, or can be configured to receive an input of a notification destination according to a first complaint risk index and a second complaint risk index, which will be described later. be.
For example, the user terminal 30 accepts direct input of the user's and/or other users' email addresses for predetermined keywords included in the subject, text, and sender's email address as the email content, or accepts group selection. Accept input. In addition, the user terminal 30 accepts direct input of the user's and/or other user's e-mail address for each numerical value at preset intervals as the first complaint risk index and the second complaint risk index, or Accepts group selection input.
The user terminal 30 transmits the received conditions and notification destination to the computer 10 .
By receiving these conditions and notification destinations, the notification destination reception module receives the input of alert notification destinations and conditions.
In this way, the notification destination receiving module can be configured to receive input of other conditions in addition to receiving input of e-mail addresses and groups.

Note that the user terminal 30 is described as accepting the input of the user's e-mail address as the notification destination, but it is not limited to the e-mail address, and the input of other contents that can identify the user, such as the user's identifier, is accepted. A configuration that accepts is also possible.

The setting module sets an alert notification destination (step S21).
The setting module sets the received notification destination as the alert notification destination.
When the setting module receives the e-mail address, it sets this e-mail address as the alert notification destination. When accepting a group, the setting module sets each e-mail address registered in this group as an alert notification destination.
Here, the setting module may be configured to set the user terminal 30 managed by the identified user as a notification destination when receiving input of contents that can identify the user other than the e-mail address.

The notification destination storage module stores the set notification destination of the alert (step S22).

The above is the notification destination setting process.
The computer 10 uses the notification destination set by the notification destination setting process described above to execute the processing described later.

[Notification Destination Change Processing Executed by Computer 10]
Based on FIG. 5, the notification destination changing process executed by the computer 10 will be described. This figure is a diagram showing a flowchart of the notification destination change processing executed by the computer 10 . This notification destination change processing is processing for changing the notification destination set by the notification destination setting processing described above.

The change acceptance module accepts a change of the alert notification destination (step S30).
The notification destination is the notification destination set by the notification destination setting process described above.
The computer 10 outputs to the user terminal 30 a predetermined input screen for receiving an input to change the notification destination. The user terminal 30 receives this input screen and displays it on its own display unit. The computer 10 accepts the input of the notification destination through a predetermined input screen displayed on the user terminal 30 .
The user terminal 30 accepts selection input of the notification destination to be changed. The user terminal 30 accepts selection input of the mail address and/or group to be changed.
Furthermore, the user terminal 30 accepts input for changing the notification destination by directly inputting a new user's e-mail address or by accepting input for selecting a new group. The method of receiving the input of the new user's mail address and/or group to be notified is the same as that of the process of step S20 described above.
The user terminal 30 transmits to the computer 10 these notification destinations for which changes have been received.
The change acceptance module accepts changes to alert notification destinations by receiving these changes to notification destinations.

It should be noted that, in step S30, the change acceptance module may be configured to accept changes of other conditions as well. Specifically, the computer 10 can change the content of the mail, input changes to the first complaint risk index and the second complaint risk index, change the notification destination set in the mail content, and change the notification set in the second complaint risk index. A configuration that accepts previous changes is also possible.
In addition, in the case where contents that can specify the user other than the e-mail address are set as the notification destination, it is also possible to accept a change in the contents.

The change module changes the setting of the alert notification destination (step S31).
The change module changes the alert notification destination set by the process of step S21 described above to the alert notification destination for which the change is accepted.

The notification destination storage module stores the changed alert notification destination (step S32).

The above is the notification destination change processing.

[Alert Notification Processing Executed by Computer 10]
Alert notification processing executed by the computer 10 will be described with reference to FIG. This figure is a diagram showing a flowchart of alert notification processing executed by the computer 10 . This alert notification process is the details of the second mail acquisition process (step S5), the second complaint risk index prediction process (step S6), and the alert notification process (step S7).

The second acquisition module acquires the second mail newly received from the trading partner (step S40).
A second acquisition module acquires, as a second email, an email newly received by the user from a business partner.
When receiving a new mail, the mail server 20 transmits the mail to the user terminal 30 managed by the user who has the recipient's mail address of this mail, and also transmits this mail to the computer 10 as the second mail.
The second acquisition module acquires the second mail by receiving this second mail.
It should be noted that the second acquisition module can also be configured to acquire, in addition to the second email, the email sent by the user to the business partner.

The unsolicited e-mail determination module determines whether the acquired second e-mail is unsolicited e-mail (step S41).
The unsolicited e-mail determination module executes processing related to filtering of existing unsolicited e-mails and determines whether the obtained second e-mail is unsolicited e-mail.
When the spam e-mail determination module determines that the acquired second e-mail is spam (step S41 YES), the computer 10 terminates this alert notification process.

On the other hand, if the spam e-mail determination module determines that the acquired second e-mail is not spam (step S41 NO), the prediction module, based on the learned model, for the newly acquired second e-mail: A second complaint risk index indicating the risk of complaints received from customers in the future is predicted (step S42).
The prediction module predicts the second complaint risk index for the acquired second mail using the learned model generated by the learning process described above. The prediction module refers to the learned model and predicts the second complaint risk index based on the obtained mail content of the second mail.
The prediction module compares the email content of the second email acquired this time with the email content in the learned model, and identifies the email content of the first email that matches or approximates the email content of the second email. The prediction module compares the subject, body, sender's email address, and recipient's email address of the first email with the subject, body, sender's email, and recipient's email of the second email, and determines whether the subject matches. The degree of matching, the number and frequency of occurrence of predetermined keywords in the text, the degree of matching of the sender's e-mail address, and the degree of matching of the recipient's e-mail address are confirmed. The prediction module identifies the first email in the learned model that best matches the email content of the second email acquired this time. If the prediction module fails to identify the first email in the trained model that matches the email content of the second email acquired this time, it identifies the first email in the learned model that is most similar to the email content of this second email. do.
The prediction module predicts the first claim risk index associated with the identified first email as the second claim risk index for the second email.
The prediction module also associates the obtained second mail with the predicted second claim risk index.
If the prediction module cannot identify the first email in the learned model that matches and approximates the email content of the second email acquired this time, the second complaint risk index of this second email is unpredictable. At this time, the prediction module preliminarily sets a second complaint risk index for a second email whose second complaint risk index is unpredictable, and thereby converts the second mail to the set second complaint risk index. A configuration for predicting as is also possible. Here, the prediction module may set the second complaint risk index of the second mail as the highest value, the lowest value, the average value, or any other numerical value.
In addition, the notification module sends an alert associated with the unpredictable second mail with the unpredictable second claim risk index to the above-described notification destination setting process or the unpredictable second claim risk index. A configuration is also possible in which notification is made to the notification destination set by notification destination change processing. At this time, the computer 10 performs re-learning processing, which will be described later, to accept the correction of the second complaint risk index of the second mail which was unpredictable. However, it is possible to predict the second claim risk index.

The first condition determination module determines whether the predicted second claim risk index satisfies the first condition (step S43).
The first condition is, for example, that the value of the second complaint risk index is 500 or more.
If the first condition determination module determines that the predicted second complaint risk index does not satisfy the first condition (step S43 NO), the first condition determination module will determine whether any future complaints from business partners in response to this e-mail. The computer 10 determines that the risk of receiving an alert is small, and terminates this alert notification process.

If the first condition determination module determines that the predicted second complaint risk index satisfies the first condition (step S43 YES), the first condition determination module will respond to this e-mail in the future and will respond to complaints from business partners in the future. Having determined that there is a possibility of receiving the claim, the second condition determination module determines whether the predicted second claim risk index further satisfies a second condition (step S44).
The second condition, for example, has a second claim risk index value higher than that of the first condition, specifically, a second claim risk index of 2,000 or more.

When the second condition determination module determines that the predicted second complaint risk index does not satisfy the second condition (step S44 NO), the notification module sends the obtained second mail and the predicted second complaint risk index is notified (step S45).
The notification module creates a first alert that associates the mail content of the obtained second mail with the predicted second complaint risk index (see FIG. 7).

[1st alert]
A first alert created by the notification module will be described with reference to FIG. This figure is a diagram schematically showing an example of the first alert created by the notification module.
The notification module accepts a message 41 pointing out the possibility of a complaint, an email display field 42 that associates the content of the second email with the second complaint risk index, and inputs from the user terminal 30 of the notification destination. A first alert 40 is created in which a confirmation icon 43 for transitioning to a management screen to be displayed is arranged at a predetermined position.
The message 41 is a character string pointing out that there is a possibility of complaints occurring in the obtained second email. It is shown. Note that the character string of this message 41 is not limited to the example described above, and its content and arrangement can be changed as appropriate as long as it points out the possibility of a complaint.
The mail display field 42 includes, as the contents of the mail, the date and time when this mail was sent to the mail server 20, the numerical value of the second claim risk index, the subject of the mail, the mail address of the sender of the mail, and the mail address of the recipient of the mail. be The figure shows the date and time of transmission, complaint risk index, title, sender's email address, and receiver's email address. Note that the content of the mail display field 42 is not limited to the example described above, and the content and arrangement thereof can be changed as appropriate as long as it is associated with the second mail and the second complaint risk index.
When the user terminal 30 displays the first alert 40, the confirmation icon 43 is used to transition to a management screen, which will be described later, by receiving an input. A configuration in which the confirmation icon 43 is not created is also possible. In this case, only the message 41 and the mail display column 42 should be arranged in the first alert 40 . Also, the user terminal 30 can be configured to directly transition to the mail screen displaying the mail contents of the second mail by accepting the input of the confirmation icon 43 .

The notification module transmits the created first alert 40 to the notification destination set by the notification destination setting process and the notification destination change process described above. The notification module transmits the first alert 40 to the user having the e-mail address set as the notification destination or to the user based on the content that can identify the user set as the notification destination.
The user terminal 30 managed by the user set as the notification destination receives the first alert 40 and displays it on its own display unit. The user terminal 30 transitions to a management screen, which will be described later, by accepting the input of the confirmation icon 43 in the displayed first alert 40 .
The notification module displays the first alert 40 on the user terminal 30 to notify the alert that associates the acquired second mail with the predicted second claim risk index.

According to the complaint occurrence prediction system 1, it is possible to predict the occurrence of complaints at the time when a new mail is received, thereby reducing the possibility of occurrence of complaints. It is possible to improve complaint handling and improve sales and repeat rate.

Returning to FIG. 6, the continuation of the alert notification process will be described.
If the second condition determination module determines that the predicted second complaint risk index satisfies the second condition (step S44 YES), the notification module determines whether the obtained second e-mail and the predicted second complaint risk index meet the second condition. are associated with each other, and an alert whose attention level has been changed is notified (step S46).
Changes in attention include, for example, making messages bold, adding highlights, adding icons, and changing the sizes and colors of characters and icons. Note that the content of the attention level change is not limited to the example described above, and any content can be used as long as it can be determined to be different from the first alert described above.
The notification module associates the content of the obtained second mail with the predicted second complaint risk index, and creates a second alert whose attention level is different from that of the first alert described above (see FIG. 8). ).

[Second alert]
A second alert created by the notification module will be described with reference to FIG. This figure is a diagram schematically showing an example of the second alert created by the notification module.
The notification module accepts a message 51 pointing out the possibility of a complaint, an email display field 52 that associates the content of the second email with the second complaint risk index, and inputs from the user terminal 30 of the notification destination. A second alert 50 is created in which a confirmation icon 53 for transitioning to the management screen to be displayed and an important icon 54 indicating that the second complaint risk index satisfies the second condition are arranged at predetermined positions.
The message 51 is a character string pointing out that there is a possibility of serious complaints occurring in the acquired second mail. A possible occurrence has been detected. Please check." is displayed. It should be noted that the change in the degree of attention in the message 51 is not limited to the example described above, and other methods may be used. Further, the character string of the message 51 is not limited to the example described above, and its content and arrangement can be changed as appropriate as long as it points out the possibility of serious complaints.
The mail display field 52 contains, as the contents of the mail, the date and time when this mail was sent to the mail server 20, the numerical value of the second complaint risk index, the subject of the mail, the mail address of the sender of the mail, and the mail address of the recipient of the mail. be The figure shows the date and time of transmission, complaint risk index, title, sender's email address, and receiver's email address. In addition, in the complaint risk index, hatching indicates that a highlight has been added to the background as a change in attention level in the mail display column 52 . It should be noted that the change in attention level is not limited to the example described above, and other methods may be used. Further, the content of the mail display column 52 is not limited to the example described above, and the content and layout thereof can be changed as appropriate as long as the second mail is associated with the second complaint risk index.
When the user terminal 30 displays the second alert 50, the confirmation icon 53 is used to transition to a management screen, which will be described later, by receiving an input. A configuration in which the confirmation icon 53 is not created is also possible. In this case, only the message 51, the mail display column 52, and the important icon 54 should be arranged in the second alert 50. FIG. Further, the user terminal 30 can be configured to directly transition to the mail screen displaying the mail contents of the second mail by accepting the input of the confirmation icon 53 .
The important icon 54 is an icon added as a change in attention level. The icon added as a change in attention level is not limited to the above example, but if it points out the possibility of a serious complaint, the content and layout can be changed as appropriate. It can be anything.
Three examples of the message 51, the mail display field 52, and the important icon 54 are described as examples of changing the degree of attention in the second alert 50. good. Also, the method of changing the degree of attention is not limited to these contents, and other methods such as changing the size and color of characters and icons may be used.

The notification module transmits the created second alert 50 to the notification destination set by the notification destination setting process and the notification destination change process described above. The notification module sends the second alert 50 to the user who has the e-mail address set as the notification destination or to the user based on the content that can identify the user set as the notification destination.
The user terminal 30 managed by the user set as the notification destination receives the second alert 50 and displays it on its own display unit. The user terminal 30 transitions to a management screen, which will be described later, by accepting the input of the confirmation icon 53 in the displayed second alert 50 .
By displaying this second alert 50 on the user terminal 30, the notification module associates the obtained second mail with the predicted second complaint risk index, and notifies the alert with the changed degree of attention.

According to the complaint occurrence prediction system 1, it is possible to predict the occurrence of complaints at the time when a new mail is received, thereby reducing the possibility of occurrence of complaints. It is possible to improve complaint handling and improve sales and repeat rate.

The above is the alert notification process.

[Mail Output Processing Executed by Computer 10]
The mail output process executed by the computer 10 will be described with reference to FIG. FIG. 1 is a diagram showing a flowchart of mail output processing executed by the computer 10. As shown in FIG. This mail output process is a process executed after the alert notification process described above.

The management screen output module outputs the management screen (step S50).
The management screen output module outputs a management screen to the user terminal 30 by accepting an input to the confirmation icon 43 or the confirmation icon 53 described above.
The user terminal 30 accepts the input of the confirmation icon 43 or the confirmation icon 53 described above, and transmits to the computer 10 information indicating that the input for these icons has been accepted. By receiving this information, the management screen output module accepts input for the confirmation icon 43 or confirmation icon 53 described above.
The management screen output module transmits the management screen to the user terminal 30 based on this information. The user terminal 30 receives this management screen and displays it on its own display unit (see FIG. 10).
The management screen output module outputs the management screen by causing the user terminal 30 to display this management screen.
It should be noted that the management screen output module can be configured to receive a predetermined input from the user terminal 30 and output the management screen to the user terminal 30 by a method other than the above-described method.

[Management screen]
A management screen output by the management screen output module will be described with reference to FIG. This figure is a diagram schematically showing an example of a management screen output by the management screen output module.
The management screen 60 is a screen that displays a list of acquired second mails. On the management screen 60, the complaint risk index, title, sender's email address, and receiver's email address of the second email are displayed as a list. In this management screen 60, the second mail for which the second alert has been notified is displayed with the degree of attention changed. As a method of changing the degree of attention, the characters are made bold, a highlight is added to the background, and an icon 61 is added.
It should be noted that the second e-mail notified of the first alert may be displayed with a different level of attention than the second e-mail notified of the second alert. That is, when changing the level of attention on the management screen 60, the display mode of the second email in which neither the first alert nor the second alert is notified is used as the reference level of attention, and the second email in which the first alert is notified is used as the standard level of attention. , the degree of attention is changed so that the display mode is more conspicuous than the reference second mail, and the display mode of the second mail in which the second alert is notified is further improved than the second mail in which the first alert is notified. Each may be displayed with a change in conspicuous prominence.
Also, the degree of attention to the second mail to which the second alert has been notified may be changed by only some of them. Also, the method of changing the degree of attention is not limited to these contents, and may be performed by a method other than these, for example, by changing the size or color of characters or icons.
By receiving a selection input for the second mail from the user terminal 30 on the management screen 60, the screen transitions to a mail screen displaying the mail contents of the second mail for which the selection input has been received (see FIG. 11).

Returning to FIG. 9, the continuation of the mail output process will be described.
The selection acceptance module accepts the selection of the second mail (step S51).
The selection reception module outputs the contents of the selected second mail to the user terminal 30 by receiving the selection input of the second mail on the management screen 60 described above.
The user terminal 30 receives the selection input of the second mail on the displayed management screen 60 and transmits information indicating that the input has been received to the computer 10 . The selection acceptance module accepts the selection of the second mail by receiving this information.

The mail output module outputs the selected second mail (step S52).
The mail output module transmits the mail contents of the selected second mail to the user terminal 30 based on the received information.
The user terminal 30 receives the mail content of this second mail and displays it as a mail screen on its own display unit (see FIG. 11).
The mail output module outputs the selected second mail by displaying the mail contents of the second mail on the user terminal 30 as a mail screen.

[Mail screen]
A mail screen output by the mail output module will be described with reference to FIG. This figure is a diagram schematically showing an example of a mail screen output by the mail output module.
The mail screen 70 is a screen for displaying the mail contents of the second mail. On the mail screen 70, the date and time of transmission, the sender's mail address, the receiver's mail address, the title, and the text are displayed as the mail contents of the second mail.
The contents of the mail displayed on the mail screen 70 are not limited to those described above, and may be only the text and the recipient's mail address, or may include other items. A second claim risk index predicted for the second mail may also be included.
In addition, based on the display mode of the mail content of the second mail in which the first alert and the second alert are not notified, the mail content of the second mail in which the first alert is notified is changed to the mail of the second mail as the reference. The degree of attention is changed so that the display mode is more conspicuous than the content, and the display mode of the second mail in which the second alert is notified is further changed than the mail content of the second mail in which the first alert is notified. Each may be displayed with a change in conspicuous prominence.
Upon receiving a predetermined input on the mail screen 70, the user terminal 30 terminates the display of the mail screen 70 and transitions to the management screen 60 again.

This is the email output process.

[Sorting Execution Processing Executed by Computer 10]
Based on FIG. 12, the rearrangement execution process executed by the computer 10 will be described. This figure is a diagram showing a flow chart of the rearrangement execution process executed by the computer 10 . This rearrangement execution process is a process related to the mail output process described above, and is a process performed at an arbitrary timing after the process of step S50 described above.

The acceptance determination module determines whether rearrangement of the second mail has been accepted (step S60).
The acceptance determination module determines whether or not an input for rearrangement has been received on the management screen 60 described above.
When the user terminal 30 receives an input to rearrange the second mail list in a predetermined order (for example, in descending order or descending order of the second claim risk index) on the displayed management screen 60, or When an input for sorting according to the ranking format is received, the received sorting information is transmitted to the computer 10.例文帳に追加
The rearrangement acceptance module accepts the rearrangement of the second mails by receiving this information.
The acceptance determination module determines whether or not the rearrangement of the second mail has been accepted by the rearrangement acceptance module, by judging whether or not the rearrangement of the second mail has been accepted.
When the reception determination module determines that the rearrangement reception module has not received a rearrangement input (step S60 NO), the computer 10 terminates the rearrangement execution process.

On the other hand, when the reception determination module determines that the rearrangement reception module has received a rearrangement input (step S60 YES), the execution module rearranges the second mails (step S61).
The execution module rearranges the second emails based on the received rearrangement content.
When receiving an input for sorting in descending order of the second complaint risk index, the execution module sorts the second mails in ascending order of the second complaint risk index (see FIG. 13). FIG. 13 shows the list of second mails displayed on the management screen 60 described above, rearranged in descending order of the second complaint risk index. Note that, when receiving an input for sorting in ascending order of the second complaint risk index, the execution module sorts the second mails in ascending order of the second complaint risk index.
Further, when receiving an input for sorting in a ranking format, the execution module ranks the second complaint risk index, sorts the second mails up to a predetermined rank in this order, and sorts the second mails with lower ranks. Exclude from the list (see FIG. 14). FIG. 14 shows the list of the second mails displayed on the management screen 60 described above, which is sorted in a ranking format. 2 Those with a low claim risk index are excluded. It should be noted that it is also possible to configure the execution modules to be displayed without excluding the execution modules that are ranked lower than a predetermined ranking.

Returning to FIG. 12, the continuation of the rearrangement execution process will be described.
The rearrangement result output module outputs the rearrangement result (step S62).
The rearrangement result output module transmits the list of the second mails after execution of the rearrangement to the user terminal 30 as the rearrangement result.
The user terminal 30 receives this rearrangement result and displays it on its display unit. At this time, the user terminal 30 displays the rearrangement result shown in FIGS. 13 and 14 as the management screen 60. FIG.
The sorting result output module outputs the sorting result by causing the user terminal 30 to display the sorting result.

The above is the sort execution process.

[Re-learning process executed by computer 10]
The relearning process executed by the computer 10 will be described with reference to FIG. 15 . This figure is a diagram showing a flowchart of the relearning process executed by the computer 10 . This re-learning process is a process that is performed after the above-described learning process, and is a process that is performed at an arbitrary timing.

The correction acceptance module accepts corrections to the predicted second claim risk index (step S70).
The user terminal 30 accepts input of the notified second complaint risk index and correction of the second complaint risk index displayed on the management screen 60 described above on a predetermined input screen. For example, the user terminal 30 accepts an input to correct the second complaint risk index upward for the second mail for which complaints were actually received, and for the second mail for which no complaints were actually received, the second Accepts input to correct the claim risk index downward.
The user terminal 30 transmits the second mail with the corrected second complaint risk index and the corrected second complaint risk index to the computer 10 .
The correction acceptance module accepts correction of the predicted second claim risk index by receiving the second mail in which the second claim risk index is corrected and the corrected second claim risk index.

The learning module associates the obtained second mail with the corrected second complaint risk index and re-learns (step S71).
The learning module uses the corrected second complaint risk index as training data, and performs supervised learning again by associating the email content of the second email with the corrected second complaint risk index with the corrected second complaint risk index. Execute.
Note that the relearning method executed by the learning module is not limited to the example described above.

The generation module updates the learned model based on the learning result (step S72).
The generation module updates the learned model based on the learning result by an existing method, similar to the process of step S14 described above.

The learned model storage module stores the updated learned model (step S73).

The above is the relearning process.
When the computer 10 newly acquires the second mail in the alert notification process described above, the computer 10 uses the learned model updated by the relearning process to execute the process of step S42 described above.

Note that the learning module performs supervised learning in which the second email for which the correction of the second complaint risk index is not accepted is associated with the second complaint risk index using the second complaint risk index as training data. is also possible. In this case, the generation module updates the learned model based on this learning result.
By doing so, the complaint occurrence prediction system 1 can further improve the accuracy of prediction of the second complaint risk index.

Although each process described above is described as a separate process, the computer 10 can also be configured to execute a combination of some or all of the above processes. Further, the computer 10 can be configured to execute the processing even at timings other than the timings described in each processing.

The means and functions described above are realized by a computer (including CPU, information processing device, and various terminals) reading and executing a predetermined program. The program may be provided, for example, from a computer via a network (SaaS: software as a service) or provided as a cloud service. Also, the program may be provided in a form recorded on a computer-readable recording medium. In this case, the computer reads the program from the recording medium, transfers it to an internal recording device or an external recording device, records it, and executes it. Alternatively, the program may be recorded in advance in a recording device (recording medium) and provided from the recording device to the computer via a communication line.

Although the embodiments of the present invention have been described above, the present invention is not limited to these embodiments described above. Moreover, the effects described in the embodiments of the present invention are merely enumerations of the most suitable effects resulting from the present invention, and the effects of the present invention are limited to those described in the embodiments of the present invention. isn't it.

(1) A claim occurrence prediction system that predicts the occurrence of a claim,
a first acquisition unit (e.g., first acquisition unit 11, first acquisition module) that acquires a plurality of first emails received from business partners;
a determination unit (e.g., determination unit 14, determination module) that determines a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of acquired first emails;
a learning unit (e.g., a learning unit 15, a learning module) that learns by associating each of the plurality of acquired first emails with the first complaint risk index determined for each;
a generation unit (eg, generation unit 16, generation module) that generates a trained model based on the learning result;
a second acquisition unit (for example, a second acquisition unit 12, a second acquisition module) that acquires a second mail newly received from the trading partner;
A prediction unit (for example, the prediction unit 17, prediction module) and
If the predicted second complaint risk index satisfies a first condition (for example, the second complaint risk index is 500 or more), an alert that associates the obtained second email with the second complaint risk index is sent. a notification unit (e.g., notification unit 13, notification module) that
Complaint occurrence prediction system.

According to the invention of (1), it is possible to improve customer dissatisfaction and complaint handling, and improve sales and repeat rate.

(2) an exclusion unit (e.g., an exclusion module) that excludes spam emails from the plurality of acquired first emails;
further comprising
The determination unit determines the first complaint risk index for each of the plurality of first emails excluding the spam emails.
The complaint occurrence prediction system according to (1).

According to the invention of (2), it is possible to improve the accuracy of determining the first claim risk index.

(3) a correction receiving unit (for example, a correction receiving module) that receives corrections to the predicted second claim risk index;
further comprising
The learning unit re-learns by associating the obtained second mail with the corrected second complaint risk index.
The complaint occurrence prediction system according to (1).

According to the invention of (3), it is possible to improve the accuracy of prediction by using the second claim risk index that is in line with reality.

(4) a first output unit (for example, a sorting result output module) that sorts and outputs the second emails in ascending or descending order of the second complaint risk index;
The complaint occurrence prediction system according to (1), further comprising:

According to the invention of (4), it becomes easy to grasp the second complaint risk index of the second mail.

(5) a second output unit (for example, a sorting result output module) that sorts and outputs the second emails according to the ranking format of the second complaint risk index;
The complaint occurrence prediction system according to (1), further comprising:

According to the invention of (5), it becomes easy to grasp the second complaint risk index of the second mail.

(6) When the predicted second complaint risk index further satisfies a second condition (for example, the second complaint risk index is 2,000 or more), the notification unit notifies the alert with the changed attention level. ,
The complaint occurrence prediction system according to (1).

According to the invention of (6), it becomes easier to grasp when the risk of a claim is higher.

(7) the notification unit notifies a preset notification destination of the alert;
The complaint occurrence prediction system according to (1).

According to the invention of (7), it is possible to notify an appropriate notification destination of an alert.

(8) a change reception unit (for example, a change reception module) that receives a change of the notification destination;
further comprising
The notification unit notifies the alert to the notification destination after the change;
The complaint occurrence prediction system according to (7).

According to the invention of (8), it is possible to notify alerts to more appropriate notification destinations.

(9) A computer-executed claim prediction method for predicting claim occurrence,
a step of obtaining a plurality of first emails received from business partners (for example, step S10);
a step of determining a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of acquired first emails (for example, step S12);
a step of learning by associating each of the plurality of first emails obtained with the first complaint risk index determined for each (for example, step S13);
a step of generating a trained model based on the learning result (for example, step S14);
a step of obtaining a second mail newly received from the trading partner (for example, step S40);
a step of predicting, based on the generated learned model, a second complaint risk index indicating the risk of complaints received from the business partner in the future for the obtained second email (for example, step S42);
if the predicted second complaint risk index satisfies a first condition, a step of notifying an alert that associates the acquired second email with the second complaint risk index (for example, step S45);
A complaint occurrence prediction method comprising:

(10) To the computer that predicts the occurrence of claims,
a step of obtaining a plurality of first emails received from business partners (for example, step S10);
determining a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of first emails obtained (for example, step S12);
learning by associating each of the plurality of first emails obtained with the first complaint risk index determined for each (for example, step S13);
a step of generating a trained model based on the learning result (for example, step S14);
a step of acquiring a second mail newly received from the trading partner (for example, step S40);
a step of predicting a second complaint risk index indicating the risk of complaints received from the business partner in the future for the acquired second mail based on the generated learned model (for example, step S42);
if the predicted second complaint risk index satisfies a first condition, a step of notifying an alert that associates the acquired second email with the second complaint risk index (for example, step S45);
A computer readable program for executing

1 Complaint Prediction System 9 Network 10 Computer 20 Mail Server 30 User Terminal 40 First Alert 41 Message 42 Mail Display Field 43 Confirmation Icon 50 Second Alert 51 Message 52 Mail Display Field 53 Confirmation Icon 54 Important Icon 60 Management Screen 61 Icon 70 mail screen

Claims (10)

1. A complaint occurrence prediction system for predicting the occurrence of a complaint,
   a first acquisition unit that acquires a plurality of first emails received from business partners;
   a determination unit that determines a first complaint risk index indicating a risk of complaints received from the business partner for each of the plurality of first emails obtained;
   a learning unit that learns by associating each of the plurality of first emails obtained with the first complaint risk index determined for each;
   a generation unit that generates a trained model based on the learning result;
   a second acquisition unit that acquires a second mail newly received from the trading partner;
   a prediction unit that predicts, based on the generated learned model, a second complaint risk index indicating the risk of complaints received from the business partner in the future for the acquired second e-mail;
   a notification unit that notifies an alert that associates the obtained second email with the second complaint risk index when the predicted second complaint risk index satisfies a first condition;
   Complaint occurrence prediction system.
2. an exclusion unit that excludes spam emails from the plurality of first emails that have been acquired;
   further comprising
   The determination unit determines the first complaint risk index for each of the plurality of first emails excluding the spam emails.
   Claim generation prediction system according to claim 1.
3. a correction receiving unit that receives a correction to the predicted second claim risk index;
   further comprising
   The learning unit re-learns by associating the obtained second mail with the corrected second complaint risk index.
   Claim generation prediction system according to claim 1.
4. a first output unit for rearranging and outputting the second mails in descending or descending order of the second complaint risk index;
   The complaint occurrence prediction system according to claim 1, further comprising:
5. a second output unit for rearranging and outputting the second mail according to the ranking format of the second complaint risk index;
   The complaint occurrence prediction system according to claim 1, further comprising:
6. If the predicted second claim risk index further satisfies a second condition, the notification unit notifies the alert with a changed attention level.
   Claim generation prediction system according to claim 1.
7. The notification unit notifies a preset notification destination of the alert;
   Claim generation prediction system according to claim 1.
8. a change reception unit that receives a change of the notification destination;
   further comprising
   The notification unit notifies the alert to the notification destination after the change;
   Claim generation prediction system according to claim 7.
9. A claim occurrence prediction method executed by a computer that predicts the occurrence of a claim,
   obtaining a plurality of first emails received from a business contact;
   determining a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of acquired first emails;
   learning by associating each of the plurality of acquired first emails with the first complaint risk index determined for each;
   generating a trained model based on the learning results;
   obtaining a second mail newly received from the trading partner;
   a step of predicting a second complaint risk index indicating the risk of complaints received from the business partner in the future for the acquired second mail based on the generated learned model;
   if the predicted second complaint risk index satisfies a first condition, sending an alert that associates the obtained second email with the second complaint risk index;
   A complaint occurrence prediction method comprising:
10. A computer that predicts the occurrence of complaints,
    obtaining a plurality of first emails received from a trading partner;
    determining a first complaint risk index indicating the risk of complaints received from the business partner for each of the plurality of acquired first emails;
    learning by associating each of the plurality of acquired first emails with the first complaint risk index determined for each;
    generating a trained model based on the learning results;
    obtaining a second mail newly received from the trading partner;
    predicting, based on the generated learned model, a second complaint risk index indicating the risk of complaints received from the business partner in the future with respect to the acquired second e-mail;
    if the predicted second complaint risk index satisfies a first condition, sending an alert that associates the obtained second email with the second complaint risk index;
    A computer readable program for executing
    
    

Images