JP2023079293A - Information processing device, information processing system, information processing method, and program - Google Patents

Abstract

To provide an information processing device, system, method and program for determining more appropriate incentive for each user.SOLUTION: In an information processing system in which a service server, a learning device and an information processing device communicate with one another through a network, the information processing device includes an acquisition part for acquiring features of a user of an object user, and a determination processing part for inputting the features of the object user to a first model and a second model, and determining an incentive to be given to the object user on the basis of an integrated score obtained by integrating a first score outputted by the first model and a second score outputted by the second model. The first model outputs a score showing the certainty that the object user performs an intended action in the case of assuming that no incentive or a first incentive is given to the user. The second model outputs a score showing the certainty that the object user performs an intended action in the case of assuming that a second incentive higher than the first incentive is given.SELECTED DRAWING: Figure 7

Description

The present invention relates to an information processing device, an information processing system, an information processing method, and a program.

Conventionally, incentives are given using a prediction model that models the purchase trends of users for each incentive generated based on the purchase history of users who have been given incentives and the background information of each user who has been given incentives. An information analysis device is disclosed that predicts the probability for optimizing an event related to purchase behavior caused by purchasing (see, for example, Patent Literature 1).

JP 2018-073349 A

However, with the conventional technology described above, there are cases where an appropriate incentive cannot be determined for each user.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an information processing apparatus, an information processing system, an information processing method, and a program capable of determining an appropriate incentive for each user. one.

One aspect of the present invention is an acquisition unit that acquires features of a target user, inputs the features of the target user into a first model and a second model, outputs a first score and the first a determination unit that determines an incentive to be given to the target user based on the integrated score obtained by integrating the second score output by the two models, wherein the first model is provided with no incentive or the user is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that the first incentive is given to the second model, the second incentive is higher than the first incentive The information processing device is a model for outputting a score indicating the probability that the target user will perform the intended action when it is assumed that the target user performs the intended action.

According to one aspect of the present invention, more appropriate incentives can be determined for each user.

1 is a diagram showing an example of a functional configuration of an information processing system 1 including an information processing device; FIG. 2 is a diagram showing an example of the functional configuration of the learning device 100; FIG. 4 is a diagram showing an example of the contents of first learning information 122. FIG. It is a figure for demonstrating the 1st admission information. 4 is a diagram showing an example of the contents of second learning information 126. FIG. It is a figure for demonstrating the 2nd admission information. 2 is a diagram showing an example of a functional configuration of an information processing device 200; FIG. 4 is a diagram showing an example of contents of user information 222. FIG. FIG. 10 is a diagram for explaining a process of obtaining a first score and a second score; FIG. FIG. 11 is a diagram for explaining processing for obtaining an integrated score; FIG. FIG. 4 is a diagram for explaining a process of determining incentives; FIG. FIG. 10 is a diagram showing the relationship between integrated scores obtained from predetermined users and admission rates; FIG. 11 is a diagram (part 1) for explaining an example of a threshold determination method; FIG. 11 is a diagram (part 2) for explaining an example of a threshold determination method; FIG. 10 is a diagram showing the total number of enrollees and incentives according to the processing of the present embodiment and the comparative example; 3 is a sequence diagram showing an example of the flow of processing executed by the information processing system 1; FIG.

Hereinafter, embodiments of an information processing apparatus, an information processing system, an information processing method, and a program according to the present invention will be described with reference to the drawings.

[overview]
An information processing apparatus according to one aspect of the present embodiment includes an acquisition unit that acquires features of a target user; and a determining unit that determines an incentive to be given to the target user based on the integrated score obtained by integrating the second score output by the second model.

User features are, for example, the user's action history, user attributes, and the like. The user's action history includes various types of information such as information indicating purchases of products or services and information indicating purchase amounts. A user's action history is a user's action history on the network. The user's action history is, for example, the user's action history in services provided via the network (purchasing history information such as purchased products and services, product prices, service prices, payment amounts, search queries, etc.) is.

An integrated score is a score obtained from the first score and the second score. An integrated score is a score obtained by statistically processing the first score and the second score. The integrated score is, for example, a score based on the difference between the first score and the second score.

The first model is a model that outputs a score indicating the probability that the target user will perform the intended action when it is assumed that there is no incentive or that the user is given the first incentive. The second model is a model that outputs a score indicating the probability that the target user will perform the intended action when it is assumed that the second incentive higher than the first incentive is given.

An intended action is an action intended by an incentive granting subject (or a predetermined person other than the user) such as an administrator of the information processing system 1 or a service provider. The intended action is, for example, any action such as applying for a predetermined service, answering a questionnaire, registering as a member, installing an application program, purchasing a product, or purchasing a service. The intended behavior may be intended by the incentive granting subject (or a predetermined person other than the user). In the following explanation, it is assumed that the intended action is membership (application) for a credit card.

[Information processing system]
FIG. 1 is a diagram showing an example of a functional configuration of an information processing system 1 including an information processing device. The information processing system 1 includes, for example, one or more user terminal devices U, one or more service servers 10 , a learning device 100 , and an information processing device 200 . One or more service servers 10, learning device 100, and information processing device 200 (or information processing device 200 and one or more service servers 10) are managed by one or the same group of administrators. . Service server 10, learning device 100, and information processing device 200 communicate with each other via network NW. The network NW includes the Internet, a LAN (Local Area Network), a WAN (Wide Area Network), a cellular network, and the like. Also, the service server 10 and the terminal device U communicate with each other via the above network NW or another network.

[Terminal device]
The terminal device U is a computer device having a communication function and a display function, such as a smart phone, a personal computer, or a tablet terminal. In the terminal device U, a processor such as a CPU (Central Processing Unit) executes a program stored in a storage device of the terminal device U to execute various processes.

[Service server]
The service server 10 is a device that provides the terminal device U with a service. The service server 10 includes, for example, a server device that provides shopping services (electronic commerce services), a server device that provides auction services, a server device that provides flea market services, restaurant reservations, hotel reservations, and travel reservations. and a server device that provides various services such as a server device that provides a search service.

[Learning device]
FIG. 2 is a diagram showing an example of the functional configuration of the learning device 100. As shown in FIG. The learning device 100 includes an acquisition unit 102, a first learning unit 104, a second learning unit 106, a providing unit 108, and a storage unit 120, for example. Acquisition unit 102, first learning unit 104, second learning unit 106, and providing unit 108 are implemented by a hardware processor such as a CPU (Central Processing Unit) executing a program (software), for example. Some or all of these components are hardware (circuit part; circuitry) or by cooperation of software and hardware. The program may be stored in advance in a storage device (a storage device with a non-transitory storage medium) such as a HDD (Hard Disk Drive) or flash memory, or may be stored in a removable storage such as a DVD or CD-ROM. It may be stored in a medium (non-transitory storage medium) and installed in the storage device by loading the storage medium into the drive device. The function of each component will be described later.

The storage unit 120 is implemented by, for example, a ROM (Read Only Memory), a flash memory, an SD card, a RAM (Random Access Memory), a HDD (Hard Disc Drive), a register, and the like. Also, part or all of the storage unit 120 may be a NAS (Network Attached Storage), an external storage server device, or the like.

The storage unit 120 stores, for example, first learning information 122, first model 124, second learning information 126, and second model 128. FIG. Details of these pieces of information will be described later. The first model 124 or the second model 128 may be, for example, a model using machine learning, a model using gradient boosting, a model using a decision tree algorithm, a support vector machine, or , a neural network, or a model using deep learning or the like. Note that the first learning information 122 is "a user who indicates that there is no incentive even if the user performs the intended action, or that the first incentive is given when the user performs the intended action." is an example of "information in which information indicating that a user performed an intended action is associated with the characteristics of the user who performed the intended action". The second learning information 126 includes “information indicating that the user has performed the intended action by indicating that a second incentive will be given when the user performs the intended action, and information indicating that the user has performed the intended action. is an example of "information associated with the features of

Acquisition unit 102 acquires first learning information 122 and second learning information 126 (described later) stored in storage unit 120, provides acquired first learning information 122 to first learning unit 104, and acquires the acquired first learning information 122. 2 provides learning information 126 to the second learning unit 106;

The first learning unit 104 learns the first learning information 122 to generate the first model 124 . FIG. 3 is a diagram showing an example of the contents of the first learning information 122. As shown in FIG. The first learning information 122 includes, for user identification information (ID), the user's action history on the network NW and the user's membership when the user is provided with credit card membership guidance without incentives. This information is associated with the first membership information indicating whether or not the membership has been completed. The user ID is, for example, a user ID used for login by the user in order to use the service provided by the service server 10 . The user ID may be a user ID using HTTP Cookie.

The user's action history is, for example, one or more action histories of an action history in an electronic commerce service, an action history in an auction service, or an action history in a search service. The user's action history includes, for example, information indicating which services are used and how much, information indicating what products or services have been purchased (purchase history information), and how much money has been used for services. It is arbitrary action history such as information indicating whether or not the product was purchased (for example, the price of the product purchased), search query used for search, and the like.

FIG. 4 is a diagram for explaining the first membership information. The service server 10 provides the terminal device U with the content shown in FIG. A user is, for example, a user who logs in to use a service provided by the service server 10 using a user ID. This content is content for recommending credit card membership, and includes credit card design and functions, credit card information such as annual membership fees, and an application button for applying for credit card membership.

For example, when the user operates an application button, a credit card application screen is displayed, and the user can apply for a credit card by inputting various information on this screen. The credit card application screen may be provided by the service server 10, or may be provided by another service server 10 cooperating with the service server 10 (the service server 10 for receiving credit card applications and managing credit card services). service server) may provide. The service server 10 manages the information about the user's operation and provides the learning device 100 with the information to be managed. The learning device 100 manages information acquired from the service server 10 as first learning information 122 .

Information on users who have applied for a credit card is positive example data (first positive example data), and information on users who have not applied for a credit card is negative example data (first negative example data). The first learning unit 104 generates a user vector by vectorizing the user's action history of the first positive example data and the first negative example data. A user vector is a distributed representation of a user's action history. A user vector indicates user characteristics. The first learning unit 104 determines that when the user vector of the first positive data is input to the model, the score output by the model approaches a predetermined value (for example, "1"), and the user vector of the first negative data is used as the model. The first model 124 is generated by performing learning so that the score output by the model when input approaches a predetermined value (for example, "zero").

As a result, the first model 124 tends to output a score of "1" or close to "1" when the user vector of the first positive data is input, and the user vector of the first negative data is input. tend to output a score of “zero” or close to “zero” when The score output by the first model 124 is an example of "a score indicating the probability that the target user will perform the intended action when it is assumed that there is no incentive or that the user is given the first incentive".

The first model 124 is such that even if the user performs the intended action, even if the user is shown that there is no incentive, when the characteristics of the user who performed the intended action are input, the intentional behavior is determined even if no incentive is given. A model trained to output a first score indicating that the user is likely to perform the intended action, or a model that indicates that the user will be given the first incentive when the intended action is performed. This is an example of a "model that has been learned to output a first score indicating that there is a high tendency to perform an intended action by being given a first incentive when characteristics of the user who performed the action are input.

The second learning unit 106 learns the second learning information 126 to generate a second model 128 . FIG. 5 is a diagram showing an example of the contents of the second learning information 126. As shown in FIG. The second learning information 126 includes, for user identification information (ID), the user's action history on the network NW and the user's membership when the user is provided with incentives and credit card membership guidance. This information is associated with the second membership information indicating whether or not the membership has been completed.

FIG. 6 is a diagram for explaining the second membership information. Differences from the first deposit information will be explained. For example, when a user who has logged in using a user ID to use a service provided by the service server 10 operates the terminal device U to purchase a product, the service server 10 performs an operation shown in FIG. The content is provided to the terminal device U. This content includes information about incentives in addition to the information included in the content shown in FIG. Incentives are, for example, subscription benefits. The incentive is, for example, that the user is given predetermined points when the user signs up for a credit card. The learning device 100 manages information acquired from the service server 10 as second learning information 126 . Incentives are not limited to points, and may be other benefits (for example, discounts, coupons, etc.) that are considered to motivate users to become members.

Information on users who have applied for a credit card is positive example data (second positive example data), and information on users who have not applied for a credit card is negative example data (second negative example data). The second learning unit 106 generates a user vector by vectorizing the user's action history of the second positive example data and the second negative example data. The second learning unit 106 determines that when the user vector of the second positive example data is input to the model, the score output by the model approaches a predetermined value (for example, “1”), and the user vector of the second negative example data is applied to the model. A second model 128 is generated by performing learning so that the score output by the model when input approaches a predetermined value (for example, "zero").

As a result, the second model 128 tends to output a score of "1" or close to "1" when the user vector of the second positive data is input, and the user vector of the second negative data is input. tend to output a score of “zero” or close to “zero” when The score output by the second model 128 is an example of "a score indicating the likelihood that the target user will perform the intended action when it is assumed that a second incentive higher than the first incentive is given."

The second model 128 is defined as "when the characteristics of the user who performed the intended action are input, the second incentive is given by indicating that the second incentive is given when the user performs the intended action. This is an example of a model that has been trained to output a second score that indicates a high tendency to perform the intended action.

The providing unit 108 provides the generated first model 124 and second model 128 to the information processing apparatus 200 .

In the above example, the user vector is generated based on the user's action history. However, in addition to (or instead of) the user's action history, a user vector that takes into account the user's attributes is used. may Further, as will be described later, the information processing apparatus 200 may similarly determine incentives for the user by using a user vector in which the user's attributes are added in addition to (or instead of) the user's action history. User attributes are, for example, various information such as sex, age, and family structure associated with the user's ID.

[Information processing device]
FIG. 7 is a diagram showing an example of the functional configuration of the information processing device 200. As shown in FIG. The information processing apparatus 200 includes, for example, an information management unit 202, a first processing unit 204, a second processing unit 206, a determination processing unit 208, and a storage unit 220. The information management unit 202, the first processing unit 204, the second processing unit 206, and the decision processing unit 208 are implemented by, for example, a hardware processor such as a CPU executing a program (software). Some or all of these components may be realized by hardware (including circuitry) such as LSI, ASIC, FPGA, GPU, etc., or by cooperation of software and hardware. good too. The program may be stored in advance in a storage device such as an HDD or flash memory (a storage device with a non-transitory storage medium), or may be stored in a removable storage medium such as a DVD or CD-ROM (non-transitory storage medium). physical storage medium), and may be installed in the storage device by mounting the storage medium in the drive device. The function of each component will be described later.

The storage unit 220 is implemented by, for example, a ROM, flash memory, SD card, RAM, HDD, register, and the like. Also, part or all of the storage unit 220 may be a NAS, an external storage server device, or the like. User information 222, a first model 224, and a second model 226 are stored in the storage unit 220, for example.

FIG. 8 is a diagram showing an example of the contents of the user information 222. As shown in FIG. The user information 222 is information provided to the service server 10 or information provided by a server device that manages the information of the service server 10 . The user information 222 is information in which user identification information and user action history are associated with each other.

The information management unit 202 acquires information provided to the service server 10 and the learning device 100, and manages the acquired information.

The first processing unit 204 acquires a first score (described later) using the first model 224 . The second processing unit 206 uses the second model 226 to obtain a second score (described later). The decision processing unit 208 acquires an integrated score by integrating the first score and the second score. The determination processing unit 208 determines an incentive to be given to the target user based on the acquired integrated score. The decision processing unit 208 decides the incentive to be given to the user, for example, based on the integrated score obtained by subtracting the first score from the second score. Determining an incentive means determining to give an incentive or determining the content of the incentive.

[Processing performed by information processing device]
FIG. 9 is a diagram for explaining the process of obtaining the first score and the second score. The information management unit 202 acquires the action history of the target user. The first processing unit 204 generates a user vector generated based on the action history of the target user.

The first processing unit 204 inputs the user vector to the first model 224 and acquires the first score output by the first model 224 . The first score is a score that indicates the possibility of joining the credit card without any incentive. For example, the closer the first score is to "1", the higher the possibility of becoming a member.

The second processing unit 206 inputs the above-described user vector to the second model 226 and acquires the second score output by the second model 226 . The second score is a score that indicates the possibility of joining a credit card with an incentive. For example, the closer the second score is to "1", the higher the possibility of becoming a member.

FIG. 10 is a diagram for explaining the process of acquiring the integrated score. The decision processing unit 208 acquires a score obtained by subtracting the first score from the second score as an integrated score. For example, if the integrated score obtained by subtracting the first score "0.5" corresponding to no incentive from the second score "0.9" corresponding to incentive is "plus 0.4", that is, the integrated score is positive , it is estimated that the target user tends to join with an incentive rather than joining without an incentive.

For example, if the integrated score obtained by subtracting the first score "0.9" corresponding to no incentive from the second score "0.5" corresponding to incentive is "minus 0.4", that is, the integrated score is negative. If , it is estimated that the target user is relatively likely to join the membership without any incentive.

FIG. 11 is a diagram for explaining the process of determining incentives. The decision processing unit 208 compares the integrated score with the threshold to decide whether to give an incentive. For example, when the integrated score is equal to or greater than the threshold Th as shown in the upper diagram of FIG. 11, the determination processing unit 208 determines to give the incentive to the target user, , it is determined not to give the incentive to the target user. A method for determining the threshold will be described later.

As a result, for example, a user determined not to give an incentive is recommended to apply for a credit card without an incentive, and a user determined to give an incentive is credited with an incentive. A recommendation to join the card is made.

Part or all of the processing described with reference to FIGS. 9 to 11 may be executed in the service server 10. For example, part or all of the functional configuration of the information processing device 200 described above may be installed in another device.

FIG. 12 is a diagram showing the relationship between integrated scores obtained from predetermined users and membership rates. The vertical axis in FIG. 12 indicates the membership rate, and the horizontal axis in FIG. 12 indicates the integrated score. Each of the plot lines in FIG. 12 is a plot line showing the relationship between the admission rate and the integrated score when there is no incentive (no incentive), and a plot line showing the relationship between the admission rate and the integrated score when incentive A is given. (Incentive A) and a plot line (Incentive B) showing the relationship between the admission rate and the integrated score when incentive B (>A) is given.

Specifically, the plot line (without incentive) is obtained from the integrated score for each user who recommended membership without incentive in the past and the information on whether or not there was membership. The admission rate when focusing on the integrated score "zero" of the plot line (no incentive) is, for example, among the users who were recommended to join without incentives in the past, the users with the integrated score of "zero" joined. percentage. Since other plot lines are similar, detailed description is omitted. In this way, it is possible to recognize the joining rate of users with the same recommendation conditions and the same integrated score from the integrated score of the plot line.

Looking at the plot line (without incentives), we can see that the enrollment rate is higher when the integrated score is low (zero or negative) than when it is high. As a result, it can be seen that users whose integrated scores are not high are likely to become members even if incentives are not given. Referring to the plotted line (incentive A) and the plotted line (incentive B), it can be seen that the membership rate is higher when the degree of incentive given is greater than when the degree of incentive is small. Furthermore, by focusing on the integrated score of each plot line, it can be understood to what degree the admission rate changes depending on the degree of incentive provision.

[Processing in which a threshold is set]
Based on the above trends, incentives are determined. For example, a threshold used by the decision processing unit 208 is determined. FIG. 13 is a diagram (part 1) for explaining an example of a threshold determination method; Differences from FIG. 12 will be described. For example, a threshold is set for the integrated score. For example, the expected number of enrollees when recommending enrollment without incentives to users with integrated scores below the threshold, and the number of enrollees when recommending enrollment with incentives to users with integrated scores above the threshold The threshold is optimized so that the sum of the number of enrollees and the number of enrollees is the desired number of enrollees or more or the maximum.

Further, when optimized, the total incentive may be optimized to meet the conditions. For example, when the incentive is to give points, the total number of points given to the user may be optimized so that the total number of new members is greater than or equal to the desired number of new members. Furthermore, various conditions may be set based on the same concept as above. The various conditions include that the total of incentives is minimized, that the membership of a user whose frequency of use of the service server 10 is equal to or higher than a predetermined degree, and the like.

For example, a threshold Th1 may be set. The threshold Th1, as shown in FIG. 13, is a threshold corresponding to an integrated score that is expected to achieve a desired enrollment rate and enrollment number even without incentives. The threshold Th1 may be, for example, an integrated score at which the admission rate of the plot line (no incentive) intersects (exceeds) the admission rate of the plot line (incentive B). By setting the threshold Th1 in this way, a high admission rate and a high number of new members can be expected without giving incentives to users who are less than the threshold Th1. As a result, it is possible to efficiently induce membership while suppressing the provision of incentives.

In the above example, granting of incentives is determined depending on whether or not the integrated score is less than the threshold Th1. Alternatively, granting of incentives and the degree of incentives to be granted are determined based on thresholds of 2 or more. may be

FIG. 14 is a diagram (part 2) for explaining an example of a threshold determination method. Differences from FIG. 13 will be described. In FIG. 14, a threshold Th2 is set. The threshold Th2 is a value corresponding to an integrated score greater than the threshold Th1. Incentive A is given to users whose integrated score is greater than or equal to threshold Th1 and less than threshold Th2, and incentive B is given to users whose integrated score is greater than or equal to threshold Th2. Also in the above case, the threshold Th1 and the threshold Th2 are set based on, for example, the total incentives to be given, the number of users that can be expected to be acquired, the expected membership rate, various constraints, and the like.

The above threshold may be derived by statistical processing performed by the information processing apparatus 200 (the determination processing unit 208), or may be provided from another apparatus. Further, for example, when the administrator's desired conditions (desired number of acquired members and constraint conditions) are input, for example, the information processing apparatus 200 may automatically derive a threshold value so as to satisfy the conditions. Further, in this case, if a specification to derive one threshold or to derive a plurality of thresholds is input, the information processing apparatus 200 may derive thresholds based on this specification. good. Furthermore, in this case, the information processing apparatus 200 may automatically determine the presence or absence of incentives and the content thereof for each category divided by thresholds.

[Comparison with Comparative Example]
FIG. 15 is a diagram showing the total number of enrollees and incentives according to the processing of the present embodiment and the comparative example. The left diagram of FIG. 15 shows the total number of enrollees and incentives by the processing of the comparative example. In the comparative example, the presence or absence and degree of incentives are changed for each route. A route is a route in a network such as what kind of content is followed.

The right diagram of FIG. 15 shows the total number of enrollees and incentives when the information processing apparatus 200 of the present embodiment is used to change the presence or absence and degree of incentives for each user. In this embodiment, as described above, users who are expected to join without giving incentives are not given incentives, and incentives are given to users who are expected to join. to give In this embodiment, by efficiently providing incentives, it is possible to efficiently acquire a larger number of members.

In particular, in this embodiment, an integrated score obtained by subtracting the first score from the second score (using the integrated score and the threshold) can be used to determine incentives for the user. For example, another comparative example different from the above is a model of the purchasing tendency of users for each incentive generated based on the purchase history of users who have been given incentives and the background information of each user who has been given incentives. "Information analysis device that predicts the probability for optimizing events related to purchase behavior caused by giving incentives using a customized prediction model" does not use the above integrated score, and provides appropriate incentives. In some cases, they were unable to make decisions or easily determine appropriate incentives. On the other hand, in this embodiment, a suitable incentive can be determined by a simple method.

[Sequence Diagram]
FIG. 16 is a sequence diagram showing an example of the flow of processing executed by the information processing system 1. As shown in FIG. First, the information processing device 200 acquires the first model 224 and the second model 226 from the learning device 100 (S10). Next, the terminal device U of the user transmits a request to the service server 10 to display a predetermined screen (for example, a screen for making payment for purchasing a product) according to the user's operation (S12).

Next, the service server 10 transmits a request for providing an integrated score to the information processing device 200 in response to obtaining the request (S14). The request of S14 includes, for example, the user's action history. Next, the information processing device 200 derives an integrated score according to the request (S16). The information processing device 200 inputs the user vector obtained from the user's action history to the first model 224 and the second model 226 as described above, and converts the first score and the second model 226 output by the first model 224 into Integrate the second scores output by to obtain an integrated score.
The information processing apparatus 200 gives an incentive to the user or determines the content of the incentive to be given to the user based on the integrated score and the threshold (S18).

Note that the request of S14 includes the user ID, and the information processing apparatus 200 may refer to the user information 222 and acquire the user's action history from the user ID.

Next, the information processing device 200 transmits information on the determined incentive to the service server 10 (S20). Next, the service server 10 determines an incentive for the user based on the transmitted incentive information, and causes the display unit of the terminal device U to display content including the determined incentive information (S22). For example, the content shown in FIG. 4 described above is provided to users without incentives, and the content shown in FIG. 6 described above is provided to users with incentives, for example.

As described above, the information processing system 1 determines an appropriate incentive for the user and provides information on the determined incentive to the user, thereby efficiently prompting the user to perform the intended action.

According to the embodiment described above, the information processing apparatus 200 inputs the features of the target user to the first model and the second model, and the first score output by the first model and the second score output by the second model are input. By determining the incentive to be given to the target user based on the integrated score obtained by integrating the scores, it is possible to determine a more appropriate incentive for each user.

As described above, the mode for carrying out the present invention has been described using the embodiments, but the present invention is not limited to such embodiments at all, and various modifications and replacements can be made without departing from the scope of the present invention. can be added.

1 -- information processing system 100 -- learning device 102 -- acquisition unit 104 -- first learning unit 106 -- second learning unit 108 -- providing unit 122 -- first learning information 124 -- first model 126 -- second learning information 128 -- second Model 200... Information processing device 202... Information management unit 204... First processing unit 206... Second processing unit 208... Decision processing unit 224... First model 226... Second model

Claims (13)

an acquisition unit that acquires characteristics of a target user;
The characteristics of the target user are input to the first model and the second model, and the integrated score obtained by integrating the first score output by the first model and the second score output by the second model a determination unit that determines an incentive to be given to the target user based on,
The first model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that there is no incentive or that the user is given the first incentive,
The second model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that a second incentive higher than the first incentive is given.
Information processing equipment. The first model indicates that there is no incentive even if the user performs the intended action, or that the first incentive is given if the user performs the intended action. is a learned model generated by learning learning information in which information indicating that the is performed the intended action is associated with the characteristics of the user who performed the intended action,
The second model includes information indicating that the user has performed the intended action by indicating that the second incentive is given when the user performs the intended action, and information indicating that the user has performed the intended action. A trained model generated by learning learning information associated with the characteristics of the user who performed
The information processing device according to claim 1 . In the first model, even if the user is shown that there is no incentive even if the user performs the intended action, the incentive is not given when the characteristics of the user who performed the intended action are input. output a first score indicating that there is a high tendency to perform the intended action even if the
The first model indicates that a first incentive is given when the user performs the intended action, and when the characteristics of the user who performed the intended action are input, the first incentive is given. Outputting a first score indicating that there is a high tendency to perform the intended action when given,
The second model indicates that a second incentive is given when the user performs the intended action, and when the characteristics of the user who performed the intended action are input, the second incentive is given. Outputting a second score indicating a high tendency to perform the intended action given
The information processing apparatus according to claim 2. The determination unit determines an incentive to be given to the user based on an integrated score obtained by subtracting the first score from the second score.
The information processing apparatus according to any one of claims 1 to 3. The features of the user are features based on the user's action history in the network,
The information processing apparatus according to any one of claims 1 to 4. The user's action history is one or more of an action history in an e-commerce service, an action history in an auction service, or an action history in a search service,
The information processing device according to claim 5 . The action history in the e-commerce service or the action history in the auction service includes information on the product purchased by the user or the amount of money required for the purchase,
The information processing device according to claim 6 . The intended action is card enrollment, questionnaire response, member registration, or application program installation.
The information processing apparatus according to any one of claims 1 to 7. Incentives are discounts, awarding of points, or the granting of coupons;
The information processing apparatus according to any one of claims 1 to 8. an information processing apparatus according to any one of claims 1 to 9;
a service server that provides services to the user in cooperation with the user's terminal device,
The information processing device provides the service server with the content of the determined incentive for the user,
The service server causes the display unit of the terminal device to display the provided content based on the incentive;
Information processing system. an acquisition unit that acquires the first learning information and the second learning information;
a first learning unit that learns the first learning information to generate a first model;
a second learning unit that learns the second learning information to generate a second model;
The first learning information indicates that there is no incentive even if the user performs the intended action, or indicates that the first incentive is given if the user performs the intended action, and the user Information in which information indicating that the intended action was performed is associated with characteristics of the user who performed the intended action,
In the first model, even if the user is shown that there is no incentive even if the user performs the intended action, the incentive is not given when the characteristics of the user who performed the intended action are input. A model trained to output a first score indicating that there is a high tendency to perform the intended action even if the
The first model indicates that a first incentive is given when the user performs the intended action, and when the characteristics of the user who performed the intended action are input, the first incentive is given. A model trained to output a first score indicating a high tendency to perform the intended action when given,
The second learning information includes information indicating that the user has performed the intended action by indicating that a second incentive will be given when the user performs the intended action, and information indicating that the user has performed the intended action. Information associated with the characteristics of the user who performed,
The second model indicates that a second incentive is given when the user performs the intended action, and when the characteristics of the user who performed the intended action are input, the second incentive is given. A model trained to output a second score that indicates a high tendency to perform the intended action when given
Information processing equipment. the computer
Get the characteristics of the target user,
The characteristics of the target user are input to the first model and the second model, and the integrated score obtained by integrating the first score output by the first model and the second score output by the second model Based on, determine the incentive to be given to the target user,
The first model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that there is no incentive or that the user is given the first incentive,
The second model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that a second incentive higher than the first incentive is given.
Information processing methods. to the computer,
Acquire the characteristics of the target user,
The characteristics of the target user are input to the first model and the second model, and the integrated score obtained by integrating the first score output by the first model and the second score output by the second model Based on, determine the incentive to be given to the target user,
The first model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that there is no incentive or that the user is given the first incentive,
The second model is a model that outputs a score indicating the likelihood that the target user will perform the intended action when it is assumed that a second incentive higher than the first incentive is given.
program.

Images