Estimation device, learning device, estimation method, and program

Abstract

To estimate the effect of a coupon with higher accuracy. [Solution] An estimation device that estimates the effectiveness of coupons given to users in electronic payment services, comprising: an acquisition unit that acquires a first feature that is a feature of the affiliated store or shop (hereinafter referred to as the affiliated store, etc.) to which the coupon is to be given, and a second feature based on the specifications of the coupon; and an inference unit that inputs the acquired first feature and second feature into a trained model that has been trained to output information indicating the effectiveness of the coupon at the affiliated store, etc. over a specified period when the first feature and second feature are input, and obtains the output of the trained model, thereby estimating the effectiveness of the coupon.

Description

[Technical Field]

[0001] The present invention relates to an estimation device, a learning device, an estimation method, and a program. [Background technology]

[0002] Previously, a technology has been disclosed in which attribute information including the user's address and the area to which the coupon applies are input into a trained model, and the output is used to estimate the effect of distributing a coupon to users within the applicable area and users outside the applicable area (Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7542780 [Non-patent literature]

[0004] [Non-Patent Document 1] Qiita webpage<https: / / qiita.com / morinota / items / 72a5a31d0b6e99cf3b26> Summary of the Invention [Problem to be solved by the invention]

[0005] The effectiveness of a coupon varies depending on the store or affiliated store where it is applied, but conventional techniques do not take this into consideration, which has resulted in the inability to accurately estimate the effectiveness of a coupon.

[0006] The present invention has been made in consideration of these circumstances, and one of its objectives is to provide an estimation device, a learning device, an estimation method, and a program that are capable of more accurately estimating the effectiveness of coupons. [Means for solving the problem]

[0007] One aspect of the present invention is an estimation device that estimates the effectiveness of a coupon given to a user in an electronic payment service, and includes: an acquisition unit that acquires a first feature that is a feature of the affiliated store or shop (hereinafter referred to as the affiliated store, etc.) to which the coupon is to be given, and a second feature based on the specifications of the coupon; and an inference unit that inputs the acquired first feature and second feature into a trained model that is trained to output information indicating the effectiveness of the coupon at the affiliated store, etc. over a predetermined period when the first feature and second feature are input, and obtains an output from the trained model, thereby estimating the effectiveness of the coupon. [Effects of the Invention]

[0008] According to one aspect of the present invention, the effectiveness of a coupon can be estimated (or can be estimated) with higher accuracy. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 illustrates basic aspects of brick-and-mortar electronic payment. [Figure 2] FIG. 1 is a diagram illustrating an example of a configuration for performing electronic payment (terminal payment) using a payment application. [Figure 3] FIG. 10 is a diagram showing an example of the contents of user information 172. [Figure 4] FIG. 10 is a diagram showing an example of the contents of affiliated store / shop information 174. [Figure 5] FIG. 10 is a diagram showing an outline of a processing flow when a user scan is performed. [Figure 6] FIG. 10 is a diagram showing an outline of the processing flow when a store scan is performed. [Figure 7] FIG. 1 is a diagram showing an example of a configuration for performing electronic payment (card payment) using a payment card. [Figure 8] FIG. 3 is a configuration diagram of an estimation device (learning device) 300. [Figure 9] FIG. 10 is a diagram illustrating an example of the content of affiliated store primary features. [Figure 10]FIG. 10 is a diagram illustrating an example of the content of a secondary feature of a user. [Figure 11] 10 is a diagram showing an example of the relationship between periods and the content of the learning process performed by the learning unit 320. FIG. [Figure 12] FIG. 10 is a diagram showing an example of the content of processing by an inference unit 330. DETAILED DESCRIPTION OF THE INVENTION

[0010] [overview] Hereinafter, with reference to the drawings, embodiments of an estimation device, a learning device, an estimation method, and a program according to the present invention will be described. The estimation device estimates the effectiveness of coupons granted to users in electronic payment services, and the learning device generates a trained model used by the estimation device. The estimation device acquires usage history of the electronic payment service from a payment server and performs the above estimation based on the acquired history. Alternatively, the estimation device may be a function of the payment server. An electronic payment service is provided to users through collaboration between an application program and a payment server, for example. In the following description, the application program is referred to as a payment app. An electronic payment service is a service that supports payments for the purchase of goods and services at a store. A store is, for example, a physical store (real store) existing in real space, but may also include a virtual store for e-commerce transactions. A virtual store may also include a store provided by an entity different from the operator of the electronic payment service. In such a case, control is performed to transition to an interface screen of the electronic payment service when making a payment for purchases at the virtual store. In an electronic payment service, a store is treated as belonging to, for example, an affiliated store (brand), and electronic payments when purchasing at a store are primarily made between the user and the affiliated store. Alternatively, electronic payment may be made between the user and the store. First, the electronic payment service will be described, and then the learning device and the estimation device will be described.

[0011] [Electronic payment methods at brick-and-mortar stores] FIG. 1 illustrates the basic aspects of brick-and-mortar electronic payments. Electronic payments are generally carried out by three parties: a medium M held by a user U, store equipment E, and a payment system S. The medium M may be a portable computer device such as a smartphone or a credit card. The store equipment E resides in a physical brick-and-mortar store (hereinafter simply referred to as the store) in real space and may include a POS device, a wireless communication device, a credit card reader, a printed code image such as a QR Code (registered trademark), or a display device displaying the code image. In brick-and-mortar electronic payments, information that can identify the user and information about the payment amount are first shared unidirectionally or bidirectionally between the medium M and the store equipment E. At this time, either the medium M or the store equipment E optically reads various information from a code image displayed by the other, provides information via near-field communication (NFC), or reads the PAN (primary account number) using a credit card reader. Then, either the medium M or the store equipment E (the party that obtains information from the other) transmits the payment information required for the payment to the payment system S via a network NW. Both the medium M and the store equipment E may send some information to the payment system S. The payment system S manages various information about the user U and performs electronic payments between the store and the user U in various ways. Electronic payments are performed using either or both of a prepaid system and a postpaid system, or by other methods. In addition, electronic payments may also include so-called online shopping, which is performed between the user's terminal device and the payment system. The network NW includes, for example, the Internet, a LAN (Local Area Network), a wireless base station, a provider device, etc. The various devices that communicate via the network NW, which will be described below, are assumed to have communication devices such as network cards and wireless communication modules.

[0012] [Configuration (Terminal Payment)] 2 is a diagram showing an example of the configuration for performing electronic payment (terminal payment) using a payment app. This electronic payment is performed mainly by a payment app 20 running on a user terminal device 10, which is one of the media M, one or more store payment terminals 30 and one or more store code images 40, which are one of the store facilities E, and a payment server 100, which constitutes part of a payment system S. The payment server 100 communicates with the user terminal device 10, the store payment terminal 30, one or more information terminals 50, and the estimation device 300 via a network NW.

[0013] The user terminal device 10 is a portable terminal device such as a smartphone or tablet. The user terminal device 10 is a computer device having at least an optical reading function, a communication function, a display function, an input acceptance function, and a program execution function. In the following description, components for realizing these functions are referred to as a camera, a communication device, a touch panel, a central processing unit (CPU), etc. In the user terminal device 10, a processor such as a CPU executes a payment application 20, which operates in cooperation with a payment server 100 to provide electronic payment services to users. The payment application 20 is installed on the user terminal device 10 from, for example, an application distribution server (not shown) and controls the camera, communication device, touch panel, etc. of the user terminal device 10. In the following description, the terms "send information to the user terminal device 10 (or receive / acquire information from the user terminal device 10)" and "send information to the payment application 20 (or receive / acquire information from the payment application 20)" may be used interchangeably, but these terms are merely different expressions and are not intended to distinguish between them.

[0014] The store payment terminal 30 is installed, for example, in a store. The store payment terminal 30 is a computer device (or a collection of these) that has at least a product price acquisition function, an optical reading function, a program execution function, and a communication function. The store payment terminal 30 includes a so-called POS (Point of Sale) device, and the POS device may have a product price acquisition function and an optical reading function.

[0015] The store code image 40 is placed in a store and is a code image such as a QR code (registered trademark) printed on a paper or plastic medium. The store code image 40 may be displayed on a display placed in the store (or on a display of a terminal device such as a smartphone or tablet terminal).

[0016] The information terminal 50 is used by the operator of the affiliated store who oversees the stores. In electronic payment services, customers who provide goods or services are treated as affiliated stores (brands), and one or more stores exist under the affiliated store. An affiliated store may operate only one store. The information terminal 50 is a smartphone, tablet terminal, personal computer, etc. An affiliated store interface 55 runs on the information terminal 50. The affiliated store interface 55 may be an affiliated store app or a web page displayed by a general-purpose browser. The affiliated store interface 55 accepts coupon settings and the like from the affiliated store operator and transmits them to the payment server 100. By executing the affiliated store interface 55, the information terminal 50 may have the function of displaying a code image corresponding to the store code image 40 or reading a code image displayed by the user terminal device 10 (in the latter case, an optical reading function is required).

[0017] The payment server 100 communicates with the credit card server 200 via a network NW. The payment server 100 has, for example, a content provider 110, an information management unit 120, a payment processing unit 130, and a storage unit 170. The components other than the storage unit 170 are realized by, for example, a hardware processor such as a CPU executing a program (software). Some or all of these components may be implemented using a large scale integration (LSI), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or the like. The program may be realized by hardware (including circuitry) such as a Gate Array (GPU) or a Graphics Processing Unit (GPU), or may be realized by a combination 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 hard disk drive (HDD) or flash memory, or may be stored in a removable storage medium (a non-transitory storage medium) such as a DVD or CD-ROM, and installed in the storage device by inserting the storage medium into a drive device.

[0018] The storage unit 170 is a HDD, flash memory, RAM (Random Access Memory), etc. The storage unit 170 may be a NAS (Network Attached Storage) device that can be accessed by the payment server 100 via a network. The storage unit 170 stores information such as user information 172 and affiliated store / shop information 174.

[0019] The content providing unit 110 has, for example, a function of a web server, and provides information (content) for displaying various screens of the electronic payment service to the user terminal device 10. The content providing unit 110 provides the content to the user terminal device 10 in the form of a web page, and provides the user terminal device 10 with parameters required for the payment application 20 to render an image.

[0020] The information management unit 120 edits, adds, deletes, etc., user information 172 and affiliated store / shop information 174, and manages them.

[0021] 3 is a diagram showing an example of the contents of user information 172. User information 172 is information in which, for example, user URL, account ID, phone number, password, registration date, charge balance, electronic money type, terminal payment method, card payment method, various history information, identity verification flag, name, address, date of birth, email address, bank account, deferred payment settings, deferred payment condition information, etc. are associated with each other. Hereinafter, a user instance (electronic payment account) in which this information is associated may be referred to as an account. In the figure, items marked with "-" indicate that they are not set.

[0022] The user URL is used for remittance processing between users. When registering for the electronic payment service, registration of a phone number and password is required. The account ID is issued to the user by the payment server 100. The registration date is the date on which the user registered for the electronic payment service (the date on which the account was created). The charge balance is information indicating the balance of electronic money that the user has set by transferring money to the account in advance. Remittance methods include depositing money into an ATM (Automatic Teller Machine) of a designated service provider (bank) or transferring money from a registered bank account. The type of electronic money is information indicating, for example, whether the electronic money can be withdrawn or can only be used for electronic payments. The terminal payment method is setting information indicating whether the user will make electronic payment using the charge balance (balance payment) or by deferred payment in terminal payment. The card payment method is setting information indicating whether the user will make electronic payment using the charge balance (balance payment) or by deferred payment in card payment. The various historical information includes charge history, which is a record of the user transferring money to the electronic payment service in advance to increase the charge balance, and payment history, which shows the details of the payments made by the user for each payment (date and time, store ID of the store where the purchase was made, affiliated store ID, payment amount, payment method, etc.).

[0023] The identity verification flag is information indicating whether or not the user has completed identity verification using an ID document. Deferred payment can be selected if identity verification has been completed, and the user with account ID "002" in the figure has not completed identity verification and can only select balance payment as the terminal payment method. The bank account is the account number of a bank account that can be used to deposit funds into the electronic payment service. Deferred payment settings is information indicating whether or not the settings have been completed to make deferred payment selectable. Deferred payment condition information is information indicating various conditions such as the deferred payment limit and the amount used for the current month.

[0024] 4 is a diagram showing an example of the contents of affiliated store / store information 174. The affiliated store / store information 174 includes, for example, a first table 174A in which an affiliated store ID and a store ID are associated with a store URL, a second table 174B in which an affiliated store ID is associated with an affiliated store name and sales amount (described above), and a third table 174C in which a store ID is associated with a store ID. In addition to this information, the affiliated store / store information 174 may also include information such as the category of the affiliated store or store, the store's location, and payment patterns.

[0025] The payment processing unit 130 performs various processes for electronic payment. There are two methods for terminal payment: a first method (user scan) and a second method (store scan), which will be explained below.

[0026] FIG. 5 shows an overview of the process flow when a user scan is performed. First, the user terminal device 10, with the payment application 20 running, reads and decodes the store code image 40 using its optical reading function (S1). The store code image 40 contains store URL information. The payment application 20 sends first payment information, including the store URL and the user's account ID, to the payment server 100 (S2). The payment server 100 searches the affiliated store / store information 174 using the affiliated store ID and store ID corresponding to the store URL, acquires information about the affiliated store name and store name (S3), and sends this to the payment application 20 (S4). The user enters the payment amount into the payment application 20 on the screen displaying the affiliated store name and store name (S5). The payment application 20 then generates second payment information including at least the payment amount and sends it to the payment server 100 (S6).

[0027] If the "Terminal Payment Method" in the user information 172 of the user is set to "Balance Payment," the payment processing unit 130 of the payment server 100 performs electronic payment based on the received second payment information (S7-1). At this time, the payment processing unit 130 performs electronic payment by, for example, decreasing the charge balance managed in association with the user ID and increasing the item value of the affiliated store's sales proceeds. The item value of the affiliated store's sales proceeds is not used as electronic money itself, for example, but rather the amount corresponding to the item value of the sales proceeds is transferred to a bank account in a cycle determined by an agreement between the affiliated store and the electronic payment service. On the other hand, if the "Terminal Payment Method" is set to "Deferred Payment," the payment processing unit 130 transmits the first payment information and the second payment information to the credit card server 200 to request electronic payment (S7-2). The credit card server 200 performs electronic payment by adding the payment amount to the user's monthly usage amount based on the received information and deducting the monthly usage amount from the user's bank account after the closing date (S7-3).

[0028] Then, the payment processing unit 130 sends a payment completion notice (information for displaying a payment completion screen) to the payment app 20 via the content providing unit 110 (S8), and the payment app 20 displays the payment completion screen (S9). When the store code image 40 is displayed on a display installed in the store, the store code image 40 may include information on the payment amount in addition to the store URL. In this case, the procedure for the user to input the payment amount is omitted, and the information on the payment amount is included in the first payment information and sent to the payment server 100. Information on the affiliated store name and store name may be included and displayed on the payment completion screen.

[0029] FIG. 6 is a diagram showing an overview of the processing flow when a store scan is performed. First, when the payment app 20 is launched, when a payment operation is performed using the payment app 20, when an automatic update timing (e.g., every minute) occurs, and at other timings, the payment app 20 sends a request to issue a one-time code to the payment server 100 (S11). The payment processing unit 130 of the payment server 100 generates a one-time code (S12) and sends it to the payment app 20 (S13). The payment app 20 displays a code image, such as a QR code or barcode, generated based on the one-time code (S14). The user holds (presents) the display surface of the user terminal device 10 over the store payment terminal 30, and the store payment terminal 30 reads and decodes the code image using its optical reading function to obtain the one-time code, etc. (S15). The store payment terminal 30 then generates payment information including the one-time code, payment amount, affiliated store ID, store ID, etc., and sends it to the payment server 100 (S16). The payment amount information is acquired in advance by reading a barcode or manually entering it.

[0030] The payment processing unit 130 of the payment server 100 identifies the user corresponding to the one-time code based on the received information, and if the "terminal payment method" in the user information 172 of the user is set to "balance payment," it performs electronic payment based on the received second payment information (S17-1). The processing content at this time is the same as the processing of S7-1 in FIG. 5. On the other hand, if the "terminal payment method" is set to "post-payment," the payment server 100 transmits the first payment information and the second payment information to the credit card server 200 to request electronic payment (S17-2). The credit card server 200 adds the payment amount to the user's monthly usage amount based on the received information, and performs electronic payment by deducting the monthly usage amount from the user's bank account after the closing date (S17-3).

[0031] Then, the payment processing unit 130 transmits a payment completion notification to the payment application 20 via the content providing unit 110 (S18), and the payment application 20 displays a payment completion screen (S19).

[0032] Note that electronic payment may be performed using only one of the above patterns. Furthermore, the "account ID" described in FIG. 2 may be other information (e.g., a phone number) that can be used as user identification information. Furthermore, issuing a one-time code may be omitted in store scanning, and the payment application 20 may display a code image generated based on the user's account ID. In this case, the payment server 100 identifies the user corresponding to the account ID instead of identifying the user corresponding to the one-time code.

[0033] It should be noted that the "post-payment" settlement may be performed within the settlement server 100, rather than being managed by the credit card server 200. In this case, the components such as the settlement card 60 and the credit card server 200 may be omitted.

[0034] [Configuration (Card Payment)] 7 is a diagram showing an example of a configuration for performing electronic payment (card payment) using a payment card. This electronic payment is performed mainly using a payment card 60, which is one of the media M, a credit card processing terminal 70, which is one of the store facilities E, a payment server 100, which constitutes part of a payment system S, and a credit card server 200. The credit card server 200 communicates with the credit card processing terminal 70 via a network NW.

[0035] The credit processing terminal 70 is installed in the store, similar to the in-store payment terminal 30. The credit processing terminal 70 includes, for example, a credit card reader and a POS device. The credit card terminal reads a personal identification number (PIN) from an inserted or held-up credit card and compares it with the PIN entered by the user. It also transmits a primary account number (PAN) read from the credit card to the credit card server 200 via the POS device. The POS device cooperates with the credit card terminal to transmit information such as the payment amount to the credit card server 200. A payment agent (acquirer) server may be interposed between the credit processing terminal 70 and the credit card server 200; however, for simplicity, the following description omits the server. The payment card 60 is, for example, similar to a commonly used credit card, with a communication chip embedded in the card substrate. The communication chip incorporates a storage medium storing the PIN and communicates with an external device via a contactor (or a wireless antenna). Alternatively, the payment card 60 may be a magnetic card. The information (messages) sent and received when using a credit card include an authorization message for authentication and a sales message for conveying the payment amount, but detailed explanations distinguishing between these will be omitted below.

[0036] The credit card server 200 communicates with the payment server 100 via a network NW. The credit card server 200 includes, for example, an information management unit 210, a credit interface 220, a payment allocation unit 230, a credit payment processing unit 240, and a memory unit 270. The components other than the memory unit 270 are implemented by, for example, a hardware processor such as a CPU executing a program (software). Some or all of these components may be implemented by hardware (including circuitry) such as an LSI, ASIC, FPGA, or GPU, or may be implemented by a combination of software and hardware. 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 (a non-transitory storage medium) such as a DVD or CD-ROM, and installed in the storage device by inserting the storage medium into a drive device. The memory unit 270 stores information such as card user information 272.

[0037] The information management unit 210 edits, adds, deletes, etc., and manages the card user information 272. The card user information 272 is information in which, for example, information unique to a user (e.g., PAN), a card payment method, and the user's account ID (used by the payment server 100) are associated with one another. The card payment method is setting information that indicates whether the user will make electronic payment using the charged balance (balance payment) or deferred payment when making a card payment.

[0038] The credit interface 220 determines whether the BIN (Bank Identification Number) in the PAN included in the message received from the credit processing terminal 70 is a code for the company, and if it is a code for the company, passes the message received from the credit processing terminal 70 to the payment allocation unit 230, and if it is not a code for the company, discards the received message.

[0039] The payment allocating unit 230 refers to the card user information 272 of the user corresponding to the message obtained from the credit interface 220, and determines whether the "card payment method" is set to "post-payment." If the "card payment method" is set to "post-payment," the payment allocating unit 230 notifies the credit interface 220 of this and passes the message obtained from the credit interface 220 to the credit payment processing unit 240. On the other hand, if the "card payment method" is set to "balance payment," the payment allocating unit 230 adds the user's account ID to the message obtained from the credit interface 220 and sends it to the payment server 100, requesting electronic payment. When requested to make electronic payment, the payment server 100 performs the same processes as S7-1 in Figure 5 and S17-1 in Figure 6.

[0040] The credit interface 220 checks the PAN and expiration date, and verifies whether the cumulative payment amount exceeds the current month's upper limit, etc. The credit payment processing unit 240 adds the payment amount to the user's monthly usage amount based on the information contained in the message obtained from the payment allocation unit 230, and performs electronic payment by deducting the monthly usage amount from the user's bank account after the closing date.

[0041] [Learning and Estimation Devices] The learning device and the estimation device will be described below. Here, it is assumed that the learning device and the estimation device are configured as an integrated unit in terms of hardware, and that the estimation device 300 has a learning unit 320 and also functions as a learning device. However, the learning device and the estimation device may be separate devices in terms of hardware.

[0042] FIG. 8 is a configuration diagram of an estimation device (learning device) 300. The estimation device 300 includes, for example, an acquisition unit 310, a learning unit 320, an inference unit 330, and a storage unit 350. The components other than the storage unit 350 are realized, for example, by a hardware processor such as a CPU executing a program (software). Some or all of these components may be realized by hardware, or may be realized by a combination of software and hardware. The program may be stored in advance in a storage device such as an HDD or flash memory (a storage device having a non-transitory storage medium), or may be stored in a removable storage medium such as a DVD or CD-ROM and installed in the storage device by inserting the storage medium into a drive device.

[0043] The storage unit 350 is a hard disk drive (HDD), flash memory, RAM, or the like. The storage unit 350 may be a NAS device accessible by the estimation device 300 via a network. The storage unit 350 stores information such as first feature amount data 352 and second feature amount data 354 at the learning stage, payment amount trend data 356, a machine learning model 358, a trained model 360, and first feature amount data 362 and second feature amount data 364 at the inference stage. The trained model 360 is stored in the storage unit 350 when the learning unit 320 completes the learning process.

[0044] [Learning stage] In the learning stage, the acquisition unit 310 acquires necessary information from the payment server 100, such as the payment history of the user information 172, coupon issuance history, and sales information for each affiliated store and shop. In the following explanation, it is assumed that the entity that issues the coupon is the affiliated store, and the explanation focuses on the affiliated store as an example of the "affiliated store, etc." in the claims. When focusing on a store, the affiliated store primary feature and the affiliated store secondary feature can be read as the store primary feature and the store secondary feature, respectively.

[0045] The acquisition unit 310 acquires, as learning data for machine learning, a first feature that is a feature of the affiliated store to which the coupon is to be issued and a second feature based on the specifications of the coupon, and stores them in the memory unit 350 as first feature data 352 and second feature data 354 in the learning stage, respectively.

[0046] The first feature amount includes, for example, a member store-specific feature amount, a member store primary feature amount, and a member store secondary feature amount. These terms are also commonly used in the inference stage. The member store-specific feature amount includes, for example, the logarithm obtained by adding 1 to the total payment amount, the number of payments, the total cashback amount, and the total commission amount over a period of interest (for example, about three months). The member store-specific feature amount may further include the region (for example, prefecture) where the member store is located, the logarithm obtained by adding 1 to the number of active users in the region, and the industry. Note that the member store-specific feature amount may be omitted from the processing.

[0047] The member store primary feature is, for example, obtained by statistically processing the transaction histories of multiple users at member stores within a specific business type. Examples of business types include restaurants, clothing, electronics retailers, and convenience stores. The member store primary feature is derived, for example, by performing matrix factorization on an evaluation matrix in which a randomly sampled predetermined number of users (e.g., several thousand) and multiple member stores within the same business type are used as axes, and information indicating transaction histories is used as elements. FIG. 9 illustrates an example of the content of the member store primary feature. In the evaluation matrix R, the vertical axis represents user identification information u, and the horizontal axis represents member store identification information b. In the evaluation matrix R, a value of 1 is assigned to each row if the user corresponding to the row has a transaction history at the corresponding member store, and a value of 0 is assigned to each row if the user has a transaction history at the corresponding member store. The acquisition unit 310 generates this evaluation matrix R and also generates a user matrix U and a member store matrix B using alternating least squares (ALS; see Non-Patent Document 1, etc.). The vector corresponding to the horizontal axis of the member store matrix B (a vertical two-dimensional vector in the figure) is used as the member store primary feature. The number of dimensions of the merchant primary features is not limited to two and can be any number. The alternating least squares method alternately minimizes two objective functions and alternately updates two parameters. Instead of the alternating least squares method, other matrix decomposition methods, such as nonnegative matrix factorization (NMF) or singular value decomposition (SVD), may be used. Note that the user primary features are also derived using the alternating least squares method, but are not shown in the figure. The elements of the evaluation matrix R are not limited to 0 or 1, but may also be values ​​that vary depending on the number of payments and the payment amount (in this case, the merchant secondary features may be omitted). Furthermore, the elements of the evaluation matrix R may be information that represents the relationship between the merchant and the user, and may be more complex information, such as whether a coupon has been obtained but not used, or whether the store is near the user's estimated address or workplace but has not been visited. The merchant primary features calculated in this way have the property that the closer the merchants are to the common users with similar tendencies, the closer their values ​​are to each other, indicating the similarity of user tendencies within an industry.

[0048] The secondary features of affiliated stores are obtained by statistically processing the secondary features of users who have a usage history, which are obtained by compressing the dimensions of the primary features of affiliated stores for each industry and then expanding them to multiple industries.

[0049] FIG. 10 is a diagram showing an example of the contents of user secondary features. The acquisition unit 310 calculates the ratio of payment amounts for each user during a period of interest for each business type for each affiliated store, and calculates a weighted sum of the affiliated store primary features using the calculated ratio as a weight. This is used as the user secondary features. The user secondary features are a tensor (matrix) of the number of businesses multiplied by the number of dimensions of the affiliated store primary features. The user secondary features are features that indicate the type of affiliated store each user tends to use for each business type.

[0050] The acquisition unit 310 then statistically processes the user secondary features of users who have a usage history for each affiliated store during the target period (for example, by calculating an average or by weighting by the payment amount and calculating a weighted sum) to derive affiliated store secondary features. Like the user secondary features, the affiliated store secondary features are a tensor (matrix) of the number of businesses multiplied by the number of dimensions of the affiliated store primary features. The affiliated store secondary features calculated in this way represent the characteristics of users who frequently use affiliated stores in a different business from the target affiliated store. For example, if the target affiliated store is a clothing store, the affiliated store secondary features include the characteristic that users who frequently use it tend to choose stores in a grocery store that offer good deals.

[0051] The second feature based on the coupon specifications may include, for example, some or all of the following: target user group (all users / new users / dormant users, etc.), whether the coupon is fixed or variable and the associated numerical values ​​(fixed amount / coupon ratio), maximum number of times, maximum period, maximum amount, minimum payment amount, start date and number of days in the period, coupon group (online payment / offline payment / across multiple affiliated stores, etc.), segment (classification of various network services linked to electronic payment), etc.

[0052] The learning unit 320 trains the parameters of the machine learning model 358 using, for example, the first feature amount and the second feature amount as learning data and the rate of increase in the target affiliated store's GMV (total payment amount using electronic payment services) over a predetermined period after the coupon having the second feature amount is issued as correct answer data, to generate the trained model 360. The rate of increase in the target affiliated store's GMV over a predetermined period is an example of "information indicating the effectiveness of the coupon."

[0053] FIG. 11 illustrates an example of the relationship between periods and the content of the learning process performed by the learning unit 320. The GMV increase rate can be calculated, for example, by using the value immediately prior to the coupon grant period as the denominator and the average value for a predetermined period immediately following the coupon grant period as the numerator. The GMV increase rate is calculated by extracting GMV trends for member stores that previously granted coupons from payment amount trend data 356. Payment amount trend data 356 is data including, for example, daily payment amount trends for each member store or shop. When the learning unit 320 receives input of the first and second feature values ​​related to a target member store that previously granted coupons, the learning unit 320 trains the parameters of the machine learning model 358 by processing such as backpropagation so as to output the GMV increase rate for the target member store over the aforementioned predetermined period. The machine learning model 358, which has undergone a predetermined number of processes, is stored in the storage unit 350 as a trained model 360.

[0054] [Inference stage] In the inference stage, the acquisition unit 310 acquires the first feature and the second feature for, for example, an affiliated store for which coupon is being considered for granting, by processing similar to that in the learning stage, and stores them in the memory unit 350 as the first feature data 362 and the second feature data 364 for the inference stage.

[0055] FIG. 12 is a diagram showing an example of the processing performed by the inference unit 330. The inference unit 330 inputs the first feature and second feature indicated by the first feature data 362 and second feature data 364 at the inference stage into the trained model 360 trained as described above, and obtains the output of the trained model 360, thereby estimating the effectiveness of a coupon. The effectiveness of a coupon is expressed, for example, by the GMV increase rate. Because the trained model 360 has been trained using various first feature values ​​related to affiliated stores as described above, it can accurately estimate the effectiveness of a coupon.

[0056] Here, if the first feature quantities, which are the features of the member store, were only member store-specific feature quantities, the second feature quantities based on the coupon specifications would have a relatively larger amount of data, and so there is a possibility that estimation accuracy would be insufficient due to data imbalance. In this regard, in this embodiment, learning and inference are performed in addition to member store primary feature quantities and member store secondary feature quantities, thereby improving estimation accuracy.

[0057] According to the embodiment described above, the effectiveness of a coupon can be estimated with higher accuracy.

[0058] The above describes the form for carrying out the present invention using an embodiment, but the present invention is not limited to such an embodiment, and various modifications and substitutions can be made within the scope that does not deviate from the gist of the present invention. [Explanation of symbols]

[0059] E. Store Facilities M medium S payment system 10 User terminal device 20. Payment App 30 Store payment terminals 40 Store code image 60 Payment Cards 70 Credit card processing terminal 100 Payment Server 130 Payment processing unit 200 Credit Card Server 300 Estimation device (learning device) 310 Acquisition Department 320 Learning Department 330 Reasoning Department 350 Storage section 358 Machine Learning Models 360 trained models

Claims

1. An estimation device that estimates the effectiveness of a coupon given to a user in an electronic payment service, an acquisition unit that acquires a first feature quantity, which is a feature quantity of the affiliated store or shop (hereinafter, "affiliated store") to which the coupon is to be issued, based on the usage history of multiple users collected independently of coupon effectiveness measurement, and a second feature quantity based on the specifications of the coupon; an inference unit that inputs the acquired first feature amount and the second feature amount into a trained model that has been trained to output information indicating the effectiveness of a coupon at a member store or the like for a predetermined period when the first feature amount and the second feature amount are input, and obtains an output of the trained model, thereby estimating the effectiveness of the coupon; An estimation device comprising:

2. the first feature amount includes a primary feature amount of the affiliated store, etc., obtained by statistically processing the usage histories of a plurality of users at the affiliated store, etc., limited to within a business type; The estimation device according to claim 1 .

3. The primary feature amount of the affiliated store, etc. is derived by performing matrix factorization on an evaluation matrix having a plurality of users and a plurality of affiliated stores, etc. in the industry as axes and information indicating usage history as elements. The estimation device according to claim 2.

4. The first feature amount further includes a secondary feature amount of an affiliated store, etc., obtained by statistically processing a secondary feature amount of a user that is obtained by expanding the primary feature amount of an affiliated store, etc., for each business type after dimensional compression of the primary feature amount of an affiliated store, etc., for a plurality of business types, for a user with a usage history. The estimation device according to claim 3.

5. The predetermined period includes a period after the coupon grant period ends. The estimation device according to claim 1 .

6. A learning device that generates a trained model for estimating the effectiveness of coupons given to users in electronic payment services, an acquisition unit that acquires a first feature quantity, which is a feature quantity of the affiliated store or shop (hereinafter, "affiliated store") to which the coupon is to be issued, based on the usage history of multiple users collected independently of coupon effectiveness measurement, and a second feature quantity based on the specifications of the coupon; a learning unit that uses the first feature amount and the second feature amount as learning data and information indicating the effectiveness of coupons at affiliated stores or the like in a predetermined period as correct answer data, and trains a machine learning model so that when the first feature amount and the second feature amount are input, the machine learning model outputs information indicating the effectiveness of coupons at affiliated stores or the like in a predetermined period; A learning device comprising:

7. An estimation device that estimates the effectiveness of a coupon given to a user in an electronic payment service, a first feature based on the usage history of a plurality of users collected independently of coupon effectiveness measurement, and a second feature based on the specifications of the coupon, the first feature being a feature of the affiliated store or shop (hereinafter, "affiliated store") to which the coupon is to be issued; The acquired first feature amount and the second feature amount are input into a trained model that has been trained to output information indicating the effectiveness of a coupon at a member store or the like for a predetermined period when the first feature amount and the second feature amount are input, and an output of the trained model is obtained, thereby estimating the effectiveness of the coupon. Estimation method.

8. An estimation device for estimating the effectiveness of a coupon given to a user in an electronic payment service, Acquire a first feature based on the usage history of multiple users collected independently of coupon effectiveness measurement, which is a feature of the affiliated store or shop (hereinafter, "affiliated store") to which the coupon is to be issued, and a second feature based on the specifications of the coupon; inputting the acquired first feature amount and the second feature amount into a trained model that has been trained to output information indicating the effectiveness of a coupon at a member store or the like for a predetermined period when the first feature amount and the second feature amount are input, and obtaining an output of the trained model, thereby estimating the effectiveness of the coupon; A program to execute.

Images