Information processor and method and program

Abstract

To evaluate the effectiveness of an operation that prompts a user to take a specified action quantitatively, thereby contributing to cost reduction for the operation.SOLUTION: An information processor comprises an effect estimation unit 21 that estimates the effect of a specified operation on a user, which is intended to prompt the user to perform a specified action, on whether the user performs the action or not, using a machine learning model that outputs a causal score indicating the effect on one or more attributes related to the user, a risk estimation unit 22 that estimates the degree of risk based on the probability that the user will not perform the action, using a machine learning model that outputs a risk indicator indicating the degree of risk based on one or more attributes of the user, and a condition output unit 24 that outputs conditions related to the operation based on the estimated effect and risk.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a technology for controlling operations for a user.

Background Art

[0002] Conventionally, for the purpose of urging payment by credit card, that is, collecting debts, operations such as making a call to a customer are performed by an operator at a call center (see Patent Document 1). Further, a reminder support technology is known in which the tone of a sentence used by an operator for response is changed according to the time zone for performing the reminder operation (see Patent Document 2).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] Conventionally, for the purpose of causing a user to execute a predetermined action (for example, transfer, deposit into an account, etc.), an operation (for example, making a call to a customer, etc.) for prompting the user to execute the predetermined action has been performed. However, while the operation of prompting the user to execute a predetermined action is effective, there is a problem that the execution of the operation incurs a cost corresponding to the amount of the operation.

[0005] In view of the above problems, an object of the present disclosure is to quantitatively evaluate the effect of an operation for prompting a user to perform a predetermined action and contribute to suppressing the cost for the operation.

Means for Solving the Problems

[0006] An example of the present disclosure is an information processing apparatus including: an effect estimation unit that estimates an effect of a predetermined operation on a user for prompting the user to execute a predetermined action, the effect being given to whether the user executes the action; and a condition output unit that outputs a condition related to the operation on the user based on the estimated effect.

[0007] Further, an example of the present disclosure is a method in which a computer executes: a teacher data acquisition step of acquiring teacher data in which a score based on a statistic related to an execution rate of a predetermined action by a user who has received a predetermined operation among a plurality of users having a predetermined attribute and a statistic related to an execution rate of the action by a user who has not received the operation among the plurality of users is defined as a score indicating an effect of the operation on the user having the attribute; and a machine learning step of creating a machine learning model based on the teacher data.

[0008] The present disclosure can be understood as an information processing apparatus, a system, a method executed by a computer, or a program to be executed by a computer. Further, the present disclosure can also be understood as a recording medium in which such a program is recorded and can be read by a computer or other devices, machines, etc. Here, the recording medium readable by a computer or the like refers to a recording medium that accumulates information such as data and programs by an electrical, magnetic, optical, mechanical, or chemical action and can be read by a computer or the like.

Advantages of the Invention

[0009] According to the present disclosure, it is possible to quantitatively evaluate the effect of an operation for prompting a user to perform a predetermined action, and contribute to suppressing the cost for the operation.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Mode for Carrying Out the Invention

[0011] Hereinafter, embodiments of an information processing apparatus, method, and program according to the present disclosure will be described with reference to the drawings. However, the embodiments described below are examples of embodiments, and do not limit the information processing apparatus, method, and program according to the present disclosure to the specific configurations described below. In implementation, specific configurations according to the implementation mode may be appropriately adopted, and various improvements and modifications may be made.

[0012] In the present embodiment, an embodiment will be described when the technology according to the present disclosure is implemented for an operation center management system for urging payment of a credit card usage amount with a delayed payment to recover a claim. However, the system to which the technology according to the present disclosure is applicable is not limited to an operation center management system for urging payment of a credit card usage amount. The technology according to the present disclosure can be widely used for technologies for controlling operations for users, and the application target of the present disclosure is not limited to the examples shown in the embodiments.

[0013] Generally, the payment of the credit card usage amount is made by methods such as being debited from the user's account on the monthly debit date or being received from the user by a specified date. However, there are cases where the payment of the credit card usage amount may not be completed by the specified date due to reasons such as insufficient balance in the user's account or the user not making the payment by the specified date. Therefore, conventionally, in order to urge the user to pay the credit card usage amount and recover the debt, operations such as making calls to customers from the operation center (call center) and sending messages have been carried out.

[0014] Generally, in order to avoid default (non - performance of debt), the operations for the user are effective, and the higher the amount of operations, the higher the debt recovery rate. However, on the other hand, the higher the amount of operations, the higher the costs such as labor costs, system usage fees, and system maintenance costs for the operations. Therefore, in the system according to the present disclosure, in view of the above - mentioned problems, a technique for suppressing the costs related to operations is adopted without reducing the debt recovery rate. In this embodiment, an example where the predetermined operation is mainly a call to the user will be described, but the content of the predetermined operation is not limited and may be various operations for prompting the user to take a predetermined action.

[0015] <Configuration of the System> FIG. 1 is a schematic diagram showing the configuration of the information processing system according to this embodiment. In the system according to this embodiment, an information processing apparatus 1, an operation center management system 3, and a credit card management system 5 are connected to be communicable with each other. An operation terminal (not shown) for performing operations according to instructions from the operation center management system 3 is installed at the operation center, and an operator operates the operation terminal to perform operations on users. A user is a credit card user who makes a payment for the credit card usage amount via a financial institution or the like, and payment history data of the credit card usage amount is notified to the operation center management system 3 via the credit card management system 5.

[0016] The information processing apparatus 1 is an information processing apparatus for outputting data for controlling operations by the operation center management system 3. The information processing apparatus 1 is a computer including a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage device 14 such as an EEPROM (Electrically Erasable and Programmable Read Only Memory) or an HDD (Hard Disk Drive), a communication unit 15 such as a NIC (Network Interface Card), and the like. However, with regard to the specific hardware configuration of the information processing apparatus 1, appropriate omission, replacement, or addition can be made according to the embodiment. Further, the information processing apparatus 1 is not limited to a device composed of a single housing. The information processing apparatus 1 may be realized by a plurality of devices using so-called cloud or distributed computing technologies.

[0017] The operation center management system 3, the credit card management system 5, and the operation terminal are all computers equipped with a CPU, ROM, RAM, storage device, communication unit, input device, output device, etc. (illustrations are omitted). Also, these systems and terminals are not limited to devices consisting of a single housing. These systems and terminals may be realized by a plurality of devices using so-called cloud or distributed computing technologies, etc.

[0018] Figure 2 is a diagram showing an outline of the functional configuration of the information processing apparatus 1 according to the present embodiment. In the information processing apparatus 1, a program recorded in the storage device 14 is read into the RAM 13 and executed by the CPU 11, and each hardware provided in the information processing apparatus 1 is controlled, whereby the information processing apparatus functions as an information processing apparatus including an effect estimation unit 21, a risk estimation unit 22, a machine learning unit 23, and a condition output unit 24. In the present embodiment and other embodiments described later, each function provided in the information processing apparatus 1 is executed by the CPU 11 which is a general-purpose processor, but a part or all of these functions may be executed by one or a plurality of dedicated processors.

[0019] The effect estimation unit 21 estimates the effect that a predetermined operation on the user for prompting the user to execute a predetermined action has on whether the user executes the action. In the present embodiment, the predetermined action is the payment of the credit card usage amount with a delayed payment. Note that the specific payment means is not limited and may be a transfer to a designated account, a payment at a designated window, or the like. Also, in the present embodiment, the predetermined operation is a reminder call for the payment of the credit card usage amount with a delayed payment. Note that the call to the user may be an automatic call using recording or mechanical voice, or a call in which an operator (human) converses with the user. And, in the present embodiment, the effect estimation unit 21 uses a machine learning model that outputs a causality score indicating the effect of the operation on the user in response to the input of one or a plurality of user attributes related to the target user, and estimates the effect of the operation.

[0020] The risk estimation unit 22 estimates a risk based on the probability that the user does not execute the predetermined action described above. The method of expressing the risk is not limited, and various indicators may be adopted. For example, the risk can be expressed using the probability of default without the debt being recovered. In this embodiment, the risk estimation unit 22 uses a machine learning model that outputs a risk indicator indicating the probability of default without the debt being recovered for the user for the input of one or more user attributes related to the target user, and estimates the risk based on the probability that the user does not execute the action. However, the risk may be estimated according to a predetermined rule, for example, without using a machine learning model. For example, the risk may be obtained by holding corresponding values in advance for each user attribute or combination of user attributes and reading out these values. In addition, indicators other than the probability of default may be adopted as the indicator indicating the risk. For example, the risk may be classified (ranked), and which class (rank) it is may be used as an indicator.

[0021] The machine learning unit 23 generates and / or updates a machine learning model used for effect estimation by the effect estimation unit 21 and a machine learning model used for risk estimation by the risk estimation unit 22. The machine learning model for effect estimation is a machine learning model that outputs a causal score indicating the degree of the effect of an operation on a target user when data of one or more user attributes related to the target user is input. Also, the machine learning model for risk estimation is a machine learning model that outputs a risk index indicating the degree of risk based on the probability that the target user will not execute an action when data of one or more user attributes related to the target user is input. The user attributes input to these machine learning models may include, for example, demographic attributes, behavioral attributes, or psychographic attributes. Here, the demographic attributes are, for example, the user's gender, family composition, age, etc., the behavioral attributes are, for example, whether or not cash withdrawal is used, whether or not installment payment is used, the deposit and withdrawal history related to a predetermined account, the business transaction history related to any product including gambling or lottery (including the online transaction history in an online marketplace, etc.), etc., and the psychographic attributes are, for example, the preference related to gambling or lottery. However, the available user attributes are not limited to the examples in this embodiment. For example, "the time required for an operation (such as making a phone call)", "the amount of credit card usage" may also be used as attributes.

[0022] In generating and / or updating a machine learning model for effect estimation, the machine learning unit 23, for each attribute of a user, based on a statistic related to the execution rate (debt recovery rate) of an action by the user who received an operation among a plurality of users having a predetermined attribute and a statistic related to the execution rate of an action by a user who did not receive an operation among the plurality of users, defines a score as a causal score indicating the effect of the operation on the user having the attribute, and creates a machine learning model based on the teacher data (data for machine learning). In the present embodiment, as an example, when the content of the operation is a call to the user, a causal score based on the difference between the statistics is calculated by the formula of "(debt recovery rate when the user is called) - (debt recovery rate when the user is not called)", and the calculated causal score is combined with the attribute data of the corresponding user and input to the machine learning unit 23 as teacher data. In the present embodiment, an example using an average value as the statistic is described. However, as the statistic, statistical indicators such as the mode or median may be used. Here, the statistic related to the execution rate of the action may be based on the past debt recovery rate within a predetermined period (for example, a predetermined month) of each user. Also, in the present embodiment, a causal score based on the difference between the statistics is calculated, but when no statistical significant difference is recognized in the difference, the causal score may be set to zero or approximately zero. Here, an existing statistical method may be adopted for determining the presence or absence of the significant difference. For example, considering the standard error and confidence interval for a set of average recovery rates of users for each month, the statistical significance of the change in the recovery rate due to the presence or absence of a call can be considered. By doing so, it becomes possible to calculate the causal score considering the variation in the average recovery rate by users within the same group.

[0023] When creating teacher data, once a user has been contacted, they can no longer be a user who has not been contacted. Therefore, for one user, only one of the recovery rates when the user is contacted and the recovery rate when the user is not contacted can be obtained. For this reason, teacher data on the effect of contact is created for each group of users with common attributes. That is, the causal score indicating the effect of contact on a group of users having a certain common attribute is, for example, obtained by dividing a plurality of users having the common attribute into a first subgroup of users who are contacted and a second subgroup of users who are not contacted, calculating the average value of the recovery rates of the debts from the first subgroup of users who are contacted and the average value of the recovery rates of the debts from the second subgroup of users who are not contacted respectively, and calculating the difference between these average values of the recovery rates based on the above-mentioned formula. For example, when the average value of the recovery rate of the debts from the first subgroup of users who are contacted is 80% and the average value of the recovery rate of the debts from the second subgroup of users who are not contacted is 70%, the causal score indicating the effect of contact on the user group is "10".

[0024] In implementing the technology according to the present disclosure, the framework for generating a machine learning model that can be adopted is, for example, based on an ensemble learning algorithm. For example, a machine learning framework (e.g., LightGBM) based on a Gradient Boosting Decision Tree (GBDT) may be adopted in the framework. In other words, the framework may adopt a machine learning framework based on a decision tree model that passes on the error between the correct answer and the predicted value between the preceding and succeeding weak learners (weak classifiers). Here, the predicted value refers to, for example, the predicted value of a causal score or a risk index. Note that, in addition to LightGBM, the framework may adopt boosting methods such as XGBoost and CatBoost. According to the framework using a decision tree, a machine learning model with relatively high performance can be generated with less effort in parameter adjustment compared to the framework using a neural network. However, the framework for generating a machine learning model that can be adopted in implementing the technology according to the present disclosure is not limited to the examples in this embodiment. For example, instead of a gradient boosting decision tree as the learner, other learners such as random forest may be adopted, or a learner not referred to as a so-called weak learner such as a neural network may be adopted. Further, especially when a learner not referred to as a so-called weak learner such as a neural network is adopted, ensemble learning may not be adopted.

[0025] FIG. 3 is a diagram schematically showing the concept of a decision tree of the machine learning model adopted in the present embodiment. When adopting a machine learning framework of gradient boosting based on the decision tree algorithm, the branching conditions of each node of the decision tree are optimized. Specifically, in a machine learning framework of gradient boosting based on the decision tree algorithm, for user groups having attributes indicated by each of two child nodes branched from one parent node, causal scores indicating the effects of operations are calculated respectively, and the branching condition of the parent node is optimized so that the difference between these causal scores becomes large (for example, so that the difference becomes maximum, or so that it becomes equal to or greater than a predetermined threshold), that is, so that the two child nodes are cleanly branched. For example, when the attribute indicated as the branching condition of the node is age, the age set as the branching threshold may be changed, or the branching condition may be changed to an attribute other than age. In this way, by recursively optimizing the branching conditions of all nodes of the decision tree, the estimation accuracy of the effect of the operation can be improved.

[0026] In generating and / or updating a machine learning model for risk estimation, the machine learning unit 23 defines, for each attribute of a user, a statistic related to the default occurrence rate of a plurality of users having a predetermined attribute (in the present embodiment, the average value. However, statistical indicators such as the mode or the median may be used, for example), as a risk indicator indicating the degree of risk of a user having the attribute, based on teacher data, and creates a machine learning model. The calculated risk indicator is combined with the attribute data of the corresponding user and input to the machine learning unit 23 as teacher data. Also, in generating or updating a machine learning model for risk estimation, the framework for generating an adoptable machine learning model is not limited, but it is the same as the generation and / or update of the machine learning model for effect estimation described above that a machine learning framework of gradient boosting based on the decision tree algorithm may be adopted.

[0027] Based on the estimated effects and risks (in this embodiment, the causal score and risk index), the conditional output unit 24 determines and outputs conditions regarding operations (hereinafter referred to as "operation conditions") to be performed on the user at the operation center. In this embodiment, the operation conditions include at least one of whether an operation needs to be performed, the number of times an operation is performed, the order of performing operations, the contact means to the user in the operation, and the content when contacting the user in the operation. Here, examples of the contact means include making a call and sending a message, and examples of the content when contacting include the content conveyed to the user during a call and the content described in the message.

[0028] FIG. 4 is a diagram showing the relationship between the effects and risks estimated in this embodiment and the operation conditions. Basically, the conditional output unit 24 outputs operation conditions such that higher priorities are given to at least some of the operations for users with higher estimated effects and some of the operations for users with higher estimated risks. Also, the conditional output unit 24 outputs operation conditions such that lower priorities are given to at least some of the operations for users with lower estimated effects and some of the operations for users with lower estimated risks.

[0029] Here, the priority is a measure indicating the degree to which an operation is preferentially performed, which is set for a user or an operation on the user. The higher the priority of a user, the higher the possibility or frequency of receiving the operation, and the lower the priority of a user, the lower the possibility or frequency of receiving the operation. Specifically, the condition output unit 24 outputs operation conditions such that, as conditions for giving a higher priority, the operation is set to be necessary to execute, the number of executions of the operation is increased, the execution order of the operation is advanced, or the contact means or contact content with the user in the operation is made to have a higher cost or effect. Also, the condition output unit 24 outputs operation conditions such that, as conditions for giving a lower priority, the operation is set to be unnecessary to execute, the number of executions of the operation is decreased, the execution order of the operation is delayed, or the contact means or contact content with the user in the operation is made to have a lower cost or effect.

[0030] In the present embodiment, the condition output unit 24 compares the estimated causal score and risk index with preset threshold values for each of the causal score and risk index, and determines and outputs operation conditions according to the comparison result. Specifically, in the example shown in FIG. 4, for the causal score, a threshold value C1 and a threshold value C2 larger than the threshold value C1 are set, and for the risk index, a threshold value R1 (first threshold value) and a threshold value R2 (second threshold value) larger than the threshold value R1 are set. Further, for the risk index, a threshold value R3 for determining whether to target the operation condition setting according to the causal score or not, and a threshold value R4 for determining a user for whom a high-priority operation condition is set regardless of the causal score are set. In the present embodiment, an example in which the same value is adopted for the threshold value R1 and the threshold value R3, and the same value is adopted for the threshold value R2 and the threshold value R4 will be described (see FIG. 4), but different values may be adopted for these threshold values respectively.

[0031] Here, cases where operation conditions with high priority are output will be described. Case 1: For at least some of the users whose causal score is equal to or higher than the threshold C2, since the effect of the operation is high, the condition output unit 24 determines and outputs operation conditions with high priority. Case 2: Also, for at least some of the users whose risk index is equal to or higher than the threshold R2, since the risk is high, the condition output unit 24 determines and outputs operation conditions with high priority. Case 3: In particular, for users whose causal score is equal to or higher than the threshold C2 and whose risk index is equal to or higher than the threshold R2 (see the area UR indicated by the broken line in FIG. 4), the condition output unit 24 may determine and output the operation conditions with the highest priority. Case 4: However, for users whose estimated risk index is lower than the threshold R3, since they are unlikely to be the default in the first place, the condition output unit 24 does not have to output operation conditions that would be given a high priority. When it is desired to enhance the cost reduction effect, for users whose risk index is lower than the threshold R3, regardless of the causal score, it is preferable to determine and output operation conditions with low priority or operation conditions with medium priority (the area UL indicated by the broken line in FIG. 4 has a causal score of C2 or higher but a risk index lower than R3, so it does not become operation conditions with high priority). For this reason, the effect estimation unit 21 may estimate the effect of the operation for users whose risk index is estimated to be equal to or higher than the threshold R3 (the third threshold), and may omit the estimation process (not estimate the effect of the operation) for users whose risk index is estimated to be lower than the threshold R3 (the third threshold).

[0032] Next, cases where operation conditions with low priority are output will be described. Case 5: For at least some of the users whose causal score is lower than the threshold C1, since the effect of the operation is low, the condition output unit 24 determines and outputs operation conditions with low priority. Case 6: Further, for at least some of the users whose risk index is less than the threshold value R1, since the risk is low, the condition output unit 24 determines and outputs operation conditions with a low priority. Case 7: In particular, for users whose causal score is less than the threshold value C1 and whose risk index is less than the threshold value R1 (see the region LL indicated by the broken line in FIG. 4), the condition output unit 24 may determine and output the operation conditions with the lowest priority. Case 8: However, for users whose estimated risk index is equal to or higher than the threshold value R4, since there is a high possibility of being default in the first place, it is not necessary to output operation conditions with a low priority. Rather, for users whose risk index is equal to or higher than the threshold value R4, regardless of the causal score, it is preferable to determine and output operation conditions with a high priority (in FIG. 4, the region LR indicated by the broken line has a causal score less than C1 but a risk index equal to or higher than R4, so it does not become operation conditions with a low priority).

[0033] From the above, in the example shown in FIG. 4, the operation priorities for high-risk users and medium-risk and high-effect users are increased, and the operation priorities for low-risk users and medium-risk and low-effect users are decreased. In the above-described example, the regions are delimited based on the threshold values, and an example in which common operation conditions are determined for users belonging to one region has been described. However, different operation conditions may be set for each user or for each user in one region. For example, even within the same region, the operation conditions may be made different so as to have a gradation according to the high or low causal score and / or risk index.

[0034] Note that the amount of operations (total number of operations or number of users targeted by operations) by the operation center management system 3 can be changed by adjusting the above-described threshold values. For example, it is possible to increase the amount of operations by decreasing at least one or more of the threshold values C1, C2, R1, and R2, and it is also possible to decrease the amount of operations by increasing at least one or more of the threshold values C1, C2, R1, and R2.

[0035] <Flow of processing> Next, the flow of processing executed by the information processing system according to the present embodiment will be described. Note that the specific content and order of processing described below are an example for implementing the present disclosure. The specific content and order of processing may be appropriately selected according to the embodiment of the present disclosure.

[0036] FIG. 5 is a flowchart showing the flow of machine learning processing according to the present embodiment. The processing shown in this flowchart is executed at the timing specified by the administrator of the operation center management system 3.

[0037] In steps S101 and S102, a machine learning model used for effect estimation is generated and / or updated. The machine learning unit 23 calculates a causal score for each of a plurality of user attributes based on user attribute data, operation history data, and payment history data of credit card usage amounts accumulated in the operation center management system 3 or the credit card management system 5 in the past, and creates teacher data including combinations of user attributes and causal scores (step S101). Here, the operation history data includes data that can determine whether an operation has been performed on each user, and the payment history data includes data that can determine whether the credit card usage amount of each user has been paid (the presence or absence of default). Then, the machine learning unit 23 inputs the created teacher data into the machine learning model, and generates or updates the machine learning model used for effect estimation by the effect estimation unit 21 (step S102). After that, the process proceeds to step S103.

[0038] In steps S103 and S104, a machine learning model used for risk estimation is generated and / or updated. The machine learning unit 23 calculates a risk index for each of a plurality of user attributes based on user attribute data, operation history data, and payment history data of credit card usage amounts accumulated in the operation center management system 3 or the credit card management system 5 in the past, and creates teacher data including combinations of user attributes and risk indexes (step S103). Then, the machine learning unit 23 inputs the created teacher data into the machine learning model, and generates or updates the machine learning model used for risk estimation by the risk estimation unit 22 (step S104). After that, the process shown in this flowchart ends.

[0039] FIG. 6 is a flowchart showing the flow of operation condition output processing according to the present embodiment. The processing shown in this flowchart is executed at a preset timing every month. More specifically, the execution timing of the processing is set to be after the payment due date of the credit card usage amount and before the scheduled operation execution date for unpaid users.

[0040] In steps S201 and S202, the effect of the operation and the risk based on the probability that the user does not execute an action are estimated. The risk estimation unit 22 inputs data of one or more user attributes related to the target user to the machine learning model generated and / or updated in step S104 for each of a plurality of users, and acquires a risk index corresponding to the user as an output from the machine learning model (step S201). Further, the effect estimation unit 21 inputs data of one or more user attributes related to the target user to the machine learning model generated and / or updated in step S102 for each of a plurality of users, and acquires a causal score corresponding to the user as an output from the machine learning model (step S202). Then, the processing proceeds to step S203.

[0041] In step S203, the operation conditions are determined and output. The condition output unit 24 determines the operation conditions based on the causal score and the risk index estimated in steps S201 and S202, and outputs them to the operation center management system 3. In the present embodiment, the condition output unit 24 specifies and outputs the operation conditions mapped in advance to the causal score and the risk index. However, the method for determining the operation conditions is not limited to the example in the present embodiment. For example, the operation conditions may include values calculated by inputting the causal score and the risk index into a predetermined function. Then, the processing shown in this flowchart ends.

[0042] When the operation conditions are output, the operation center management system 3 manages the operations for the target users according to the operation conditions, and the operation terminal executes the operations according to the instructions output by the operation center management system 3.

[0043] <Effect> According to this embodiment, by setting the priority of the operation conditions according to the effects and risks of the operations for each user and suppressing the operations for users with low effects and risks, it is possible to suppress the costs related to the operations without reducing the recovery rate of the claims. That is, according to the present disclosure, it is possible to suppress the costs for the operations without reducing the effects of the operations that prompt the users to take a predetermined action. Further, according to this embodiment, by increasing the operations for users with high effects and risks, it is also expected to increase the recovery rate of the claims while suppressing the costs.

[0044] <Variations of the operation condition output process> In the embodiment described above, the flow of the operation condition output process has been schematically described with reference to FIG. 6. More specifically, the operation condition output process may be processed as follows.

[0045] FIG. 7 is a flowchart showing the flow of the operation condition output process when the determination methods of Cases 1 to 4 described with reference to FIG. 4 are adopted in this embodiment. According to the example shown in FIG. 7, when the risk index calculated (step S301) by the risk estimation unit 22 for a user having a certain attribute is less than the threshold value R3 (the third threshold value) (NO in step S302), the calculation of the causal score by the effect estimation unit 21 is omitted, and it can be seen that the operation conditions with low (or medium) priority are determined and output (step S303).

[0046] When the risk index calculated for the user is equal to or higher than the threshold value R3 (third threshold value) (YES in step S302), a causal score is calculated (step S304). When the causal score is equal to or higher than the threshold value C2 and the risk index is equal to or higher than the threshold value R2 (YES in step S305), the operation condition with the highest priority is determined and output (step S306). When the causal score is equal to or higher than the threshold value C2 or the risk index is equal to or higher than the threshold value R2 (YES in step S307), an operation condition with high priority is determined and output (step S308). When the causal score is less than the threshold value C2 and less than the threshold value R2, an operation condition with medium priority is determined and output (step S309).

[0047] FIG. 8 is a flowchart showing the flow of operation condition output processing when the determination method for cases 5 to 8 described with reference to FIG. 4 in the present embodiment is adopted. According to the example shown in FIG. 8, when the risk index calculated by the risk estimation unit 22 for a user having a certain attribute is equal to or higher than the threshold value R4 (NO in step S402), the calculation of the causal score by the effect estimation unit 21 is omitted, and it can be seen that an operation condition with high (or medium) priority is determined and output (step S403).

[0048] When the risk index calculated for the user is less than the threshold value R4 (YES in step S402), a causal score is calculated (step S404). When the causal score is less than the threshold value C1 and the risk index is less than the threshold value R1 (YES in step S405), the operation condition with the lowest priority is determined and output (step S406). When the causal score is less than the threshold value C1 or the risk index is less than the threshold value R1 (YES in step S407), an operation condition with low priority is determined and output (step S408). When the causal score is equal to or higher than the threshold value C1 and equal to or higher than the threshold value R1, an operation condition with medium priority is determined and output (step S409).

[0049] <Other variations> In the above-described embodiments, an example in which the operation for the user is a phone call has been described. However, the type of operation for the user is not limited to a phone call. For example, message transmission may be adopted as the type of operation for the user. Note that the means for message transmission is not limited here, and an e-mail system, a short message service (SMS), or a message transmission / reception service of a social networking service (SNS) may be used.

[0050] Also, in the above-described embodiments, an example in which the operation conditions are determined based on the estimated effects of one type of operation (phone call) has been described. However, the operation conditions may be determined based on the estimated effects for each of a plurality of types of operations (for example, phone call and message transmission). In this case, the effect estimation unit 21 estimates the first effect on whether the user executes an action given by the first operation for the user (for example, a phone call) for prompting the user to execute a predetermined action, and the second effect on whether the user executes an action given by the second operation for the user (for example, message transmission) for prompting the user to execute a predetermined action. The condition output unit 24 outputs the operation conditions for the user based on the estimated first effect and second effect.

[0051] In this case, the machine learning model for estimating the effects of operations is also generated and updated for each type of operation. For example, when the first operation is a phone call and the second operation is message transmission, a machine learning model for estimating the effects of phone calls and a machine learning model for estimating the effects of message transmission may be generated and updated.

[0052] Furthermore, when operation conditions are determined based on the estimated effects of multiple types of operations, an operation type with a high effect on the user may be selected from the multiple types of operations. In this case, the condition output unit 24 outputs operation conditions including whether to perform the first operation or the second operation on the user based on the estimated first effect and second effect. More specifically, by comparing the first effect (causal score related to the first operation) and the second effect (causal score related to the second operation) obtained for the target user, the operation type with the higher causal score can be selected as the operation type with a high effect on the user.

[0053] Also, in the above-described embodiment, an example in which two axes, namely, a causal score indicating the effect of an operation and a risk index indicating risk, are used as evaluation axes for determining operation conditions has been described. However, in the technology according to the present disclosure, it is sufficient that the evaluation axes for determining operation conditions include at least the effect of the operation, and other evaluation axes may be adopted, or three or more evaluation axes may be adopted. For example, (1) a third index other than the risk index may be adopted as an index combined with the causal score, (2) a third index may be adopted in addition to the causal score and the risk index, or (3) three axes of the causal score of power supply, the causal score of message transmission, and the risk index may be adopted.

[0054] In the above-described embodiment, an example was described in which the difference between the action execution rate of the sub-user group that received the operation and the action execution rate of the sub-user group that did not receive the operation was used as the causal score. However, as the causal score, those calculated by other methods may be used. For example, in generating and / or updating a machine learning model for effect estimation, the machine learning unit 23, for each attribute of a user, based on a statistic related to the execution rate of an action by a user who has taken a predetermined reaction to an operation among a plurality of users having a predetermined attribute and a statistic related to the execution rate of an action by a user who has not taken the predetermined reaction among the plurality of users, a score may be defined as the causal score related to the user having the attribute, and a machine learning model may be created based on the teacher data. As an example, in this variation, the causal score is calculated by the formula "(recovery rate of claims when the user takes a predetermined reaction) - (recovery rate of claims when the user does not take a predetermined reaction)". In this case, the condition output unit 24 may output conditions related to the operation based on the presence or absence and content of the reaction by the user and the causal score corresponding to the reaction. At this time, the adjustment of the priority of the operation conditions output by the condition output unit 24 may be performed using the relationship between the causal score and the priority described with reference to FIG. 4.

[0055] Here, the predetermined reaction is, for example, a user's response such as a dial push for a call, or a call with an operator accompanying a callback from the user for a call, a reply to a message, a read receipt of a message, etc. Further, as the content of the reaction, a positive answer to a payment, the presence or absence of an answer to the payment due date, etc. may be considered. When the reaction by the user is a voice reaction, it is also possible to determine the user's emotion, etc. based on the user's voice and determine whether the reaction is positive. Also, the priority of the next operation conditions may be adjusted based on the user's emotion, etc. determined based on the voice.

[0056] Also, the priority of the operation conditions may be adjusted based on elements other than those described above. For example, for users whose payment settings for credit card use are installment payments or split payments, the priority of the operation conditions may be increased compared to users who make lump-sum payments. For users whose credit card use includes cashing, the priority of the operation conditions may be increased compared to users who do not include cashing. Further, it is also possible to adjust the priority of the operation conditions according to the transaction data of the target user other than credit card transactions (for example, the balance data of the credit card usage debit account, the transaction history data in affiliated banks, etc.). Here, the condition output unit 24 may adjust various thresholds corresponding to various scores shown in FIG. 4 based on, for example, the usage conditions of the credit card.

Description of Signs

[0057] 1 Information processing apparatus

Claims

1. An effect estimation unit that estimates, using a machine learning model that outputs a causal score indicating an effect on the execution of a predetermined operation for a user to prompt the user to execute a predetermined action, the effect given to whether the user executes the action with respect to the input of one or more attributes related to the user; A risk estimation unit that estimates, using a machine learning model that outputs a risk index indicating the degree of risk based on the probability that the user does not execute the action, the degree of risk with respect to the input of one or more attributes related to the user; A condition output unit that outputs conditions related to the operation based on the estimated effect and the risk; An information processing apparatus comprising the above.

2. The condition output unit outputs conditions such that a higher priority is given to at least a part of the operation for a user with a higher estimated effect and the operation for a user with a higher estimated risk. The information processing apparatus according to Claim 1.

3. The condition output unit outputs conditions such that a lower priority is given to at least a part of the operation for a user with a lower estimated effect and the operation for a user with a lower estimated risk. The information processing apparatus according to Claim 1 or 2.

4. The condition output unit does not output conditions such that a higher priority is given to the operation for a user whose estimated risk is lower than a first threshold. The information processing apparatus according to any one of Claims 1 to 3.

5. The condition output unit does not output conditions such that a lower priority is given to the operation for a user whose estimated risk is higher than a second threshold. The information processing apparatus according to any one of Claims 1 to 4.

6. The effect estimation unit estimates the effect of the operation for a user for whom the risk is estimated to be equal to or higher than a third threshold, and does not estimate the effect of the operation for a user for whom the risk is estimated to be lower than the third threshold. The information processing apparatus according to any one of Claims 1 to 5. **Claim 7**: The conditions regarding the operation are at least any one of the necessity of executing the operation, the number of times of executing the operation, the execution order of the operation, the contact means to the user in the operation, and the content when contacting the user in the operation. The information processing apparatus according to any one of claims 1 to 6. **Claim 8**: The effect estimation unit estimates a first effect on the user for a first operation on the user to prompt the execution of a predetermined action, regarding whether the user will execute the action, and a second effect on the user for a second operation on the user to prompt the execution of the predetermined action, regarding whether the user will execute the action. The condition output unit outputs the conditions regarding the operation on the user based on the estimated first effect and second effect. The information processing apparatus according to any one of claims 1 to 7. **Claim 9**: The condition output unit outputs the conditions including whether to perform the first operation or the second operation on the user based on the estimated first effect and second effect. The information processing apparatus according to claim 8. **Claim 10** A computer performs an effect estimation step of estimating, using a machine learning model that outputs a causal score indicating the effect of a predetermined operation on the user to prompt the execution of a predetermined action, regarding whether the user will execute the action, for an input of one or more attributes related to the user; a risk estimation step of estimating, using a machine learning model that outputs a risk index indicating the degree of risk based on the probability that the user will not execute the action, for an input of one or more attributes related to the user; and a condition output step of outputting the conditions regarding the operation based on the estimated effect and risk. A method for execution. **Claim 11** A computer is configured as an effect estimation unit that estimates, using a machine learning model that outputs a causal score indicating the effect of a predetermined operation on the user to prompt the execution of a predetermined action, regarding whether the user will execute the action, for an input of one or more attributes related to the user. A risk estimation unit that estimates the degree of risk based on the probability that the user does not execute the action, using a machine learning model that outputs a risk index indicating the degree of risk for an input of one or more attributes related to the user; A condition output unit that outputs conditions regarding the operation based on the estimated effect and the risk; A program that functions as.

Images