Information processing device, information processing method, and information processing program

Abstract

To efficiently select products to be advertised. [Solution] An information processing device acquires first data including the number of impressions and clicks for an advertising page for each of a plurality of products during a specified period, acquires second data including purchase history for products advertised on the advertising pages clicked during the specified period, uses the first data and the second data to calculate a product quality score that represents the balance between CVR (Conversion Rate) and CTR (Click Through Rate) for each of the plurality of products during the specified period, and selects a specified number of products to be advertised from the plurality of products based on the product quality score.

Description

[Technical Field]

[0001] The present invention relates to a technique for selecting products to be advertised. [Background technology]

[0002] In recent years, electronic commerce (E-commerce), which involves selling products online using the Internet, has become increasingly common. For example, such E-commerce is conducted via a marketplace (e.g., an E-commerce site such as a marketplace-type E-commerce site) where multiple stores sell products by listing them online. By accessing the marketplace from a mobile device such as a personal computer (PC) or smartphone, users can browse and purchase desired products without having to physically visit multiple stores or worrying about time.

[0003] Merchants (i.e., product sellers) selling products on marketplaces distribute product advertisements using various online advertising platforms to increase the conversion rate (CVR) for the products they sell. CVR is a numerical value that indicates the ratio of the number of conversions, such as purchases or contracts, that a merchant expects to achieve to the number of accesses (also referred to as the number of visits or sessions) to a web page (here, a product page for selling products). An online advertising platform is a platform for online (i.e., on a web page) advertisement placement and distribution and advertising analysis services, and is provided by web services such as Google, Facebook, and Yahoo. Merchants can use the online advertising platform to distribute advertisements for products that may match the interests of users (i.e., web page viewers) through various channels provided by the web service (e.g., video site channels, news site channels, and hobby and preference site channels).

[0004] When delivering product advertisements using advertising pages on online advertising platforms (e.g., pages allocated to advertisements that occupy at least a portion of the web page viewed by a user), delivering advertisements for products with a high CVR is effective in terms of merchant profits. Meanwhile, since operators of online advertising platforms charge merchants based on the number of clicks on their product advertising pages (i.e., they charge advertising fees), delivering advertisements for products with a high CTR (Click-Through Rate) is effective in terms of the operator's profits. CTR is a numerical value indicating the ratio of the number of clicks to the number of impressions on a web page (i.e., the number of times a web page is displayed). For merchants, if both the CVR and CTR are high, advertising costs will be high relative to the profits from product sales, resulting in reduced overall revenues. For this reason, there is a demand for technology that creates advertisements that take CVR and CTR into consideration. A technology for creating advertisements that take CVR and CTR into consideration is disclosed, for example, in Patent Document 1. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-168297 Summary of the Invention [Problem to be solved by the invention]

[0006] The above-mentioned document discloses a technique for creating advertisements that take CVR and CTR into consideration. However, no mechanism has been proposed to efficiently select one or more products to be advertised from among multiple candidate products to be advertised, taking into consideration the balance between CVR and CTR.

[0007] The present invention has been made in view of the above-mentioned problems, and has an object to provide a technique for efficiently selecting products to be advertised. [Means for solving the problem]

[0008] In order to solve the above problem, one aspect of an information processing device according to the present invention includes a first data acquisition unit that acquires first data including the number of impressions and the number of clicks for an advertising page of each of a plurality of products during a predetermined period of time; a second data acquisition unit that acquires second data including purchase history for products advertised on the advertising page clicked during the predetermined period of time; a calculation unit that uses the first data and the second data to calculate a product quality score that represents the level of balance between CVR (Conversion Rate) and CTR (Click Through Rate) for each of the plurality of products during the predetermined period of time; and a selection unit that selects a predetermined number of products to be advertised from the plurality of products based on the product quality score.

[0009] In order to solve the above problem, one aspect of an information processing method according to the present invention includes acquiring first data including the number of impressions and the number of clicks for an advertising page of each of a plurality of products during a predetermined period of time, acquiring second data including purchase history for products advertised on the advertising page clicked during the predetermined period of time, calculating a product quality score using the first data and the second data that represents the level of balance between CVR (Conversion Rate) and CTR (Click Through Rate) for each of the plurality of products during the predetermined period of time, and selecting a predetermined number of products to be advertised from the plurality of products based on the product quality score.

[0010] In order to solve the above problem, one aspect of the information processing program according to the present invention is an information processing program for causing a computer to execute information processing, the program causing a computer to execute processes including: a first data acquisition process for acquiring first data including the number of impressions and the number of clicks for an advertising page of each of a plurality of products during a predetermined period; a second data acquisition process for acquiring second data including purchase history for products advertised on the advertising page clicked during the predetermined period; a calculation process for calculating a product quality score using the first data and the second data, which represents the level of balance between CVR (Conversion Rate) and CTR (Click Through Rate) for each of the plurality of products during the predetermined period; and a selection process for selecting a predetermined number of products to be advertised from the plurality of products based on the product quality score. [Effects of the Invention]

[0011] According to the present invention, it is possible to efficiently select products to be advertised. The above-mentioned objects, aspects, and advantages of the present invention, as well as other objects, aspects, and advantages of the present invention not described above, will be understood by those skilled in the art from the following detailed description of the invention by referring to the accompanying drawings and the claims. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 shows an example of the configuration of an information processing system according to an embodiment. [Figure 2] FIG. 2 is a conceptual diagram showing the overall processing by the product list management device, the marketplace operation device, and the advertisement distribution device. [Figure 3] FIG. 3 illustrates an example of a functional configuration of a product list management device according to an embodiment. [Figure 4] FIG. 4 illustrates an example of a hardware configuration of a product list management device according to an embodiment. [Figure 5] FIG. 5 shows a flowchart of the process executed by the product list management device according to the embodiment. [Figure 6] FIG. 6 shows an example of a product quality score. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, with reference to the accompanying drawings, an embodiment for carrying out the present invention will be described in detail. Among the components disclosed below, those having the same function will be given the same reference numerals, and their description will be omitted. Note that the embodiment disclosed below is an example of a means for realizing the present invention, and should be appropriately modified or changed depending on the configuration of the device to which the present invention is applied and various conditions. The present invention is not limited to the following embodiment. Furthermore, not all of the combinations of features described in the present embodiment are necessarily essential to the solution of the present invention.

[0014] [Configuration of information processing system] FIG. 1 shows an example configuration of an information processing system 1 according to this embodiment. The information processing system 1 includes a product list management device 10, a marketplace operation device 11, an advertisement distribution device 12, and a user device 13. The product list management device 10, the marketplace operation device 11, the advertisement distribution device 12, and the user device 13 function as information processing devices configured to be able to communicate with each other via a network 14. The network 14 may include the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a mobile communication network, and the like. Although FIG. 1 illustrates one user device 13, the information processing system 1 includes multiple user devices (not shown) having the same functions as the user device 13, and the multiple user devices are configured to be able to communicate with the product list management device 10, the marketplace operation device 11, and the advertisement distribution device 12 via the network 14. The user device 13 is operated by a user 15. In the present disclosure, the terms "user device" and "user" may be understood to be synonymous. The product list management device 10 and the advertisement distribution device 12 may be configured as a single device. The term "product" may be understood as a term that includes at least information that can identify the product.

[0015] The marketplace operating device 11 is a server device that operates and provides a marketplace (e.g., an e-commerce site such as a marketplace-type e-commerce site) where multiple stores sell products by listing them online. For example, the marketplace operating device 11 operates an e-commerce mall that operates a shopping mall via sales pages (web pages) of products sold by multiple merchants. The marketplace operating device 11 can accept access from user devices 13 via a network 14 and provide various services related to shopping in the marketplace to users 15. For example, when users 15 access the marketplace and perform actions such as purchasing or viewing on a product page of a given product (i.e., an online product sales page; the same applies below), the marketplace operating device 11 provides users 15 with services related to the product. Note that the marketplace operating device 11 is not limited to a server device and may be realized by a mainframe or the like.

[0016] The advertisement distribution device 12 is a server device that provides an online advertisement platform (hereinafter referred to as the advertisement platform) and distributes advertisements. The advertisement platform is a platform for online advertisement posting and distribution and advertisement analysis services. The advertisement distribution device 12 can post (i.e., upload) and distribute advertisements using banners or the like on multiple online channels (e.g., a video site channel, a news site channel, a hobby and preference site channel, an online map channel, and an email channel). Merchants of the marketplace provided by the marketplace operating device 11 can distribute product advertisements using the advertisement platform provided by the advertisement distribution device 12. When the user 15 clicks on an advertisement page (e.g., a page allocated to an advertisement that occupies at least a portion of the area of ​​a web page viewed by the user 15; the same applies hereinafter) of any product distributed on an online channel (hereinafter referred to as a channel) displayed on the user device 13, the page displayed on the user device 13 moves to a landing page corresponding to the advertisement. For example, when the user 15 clicks on an advertisement page for a product sold by a merchant in the marketplace provided by the marketplace operating device 11, the page displayed on the user device 13 switches to the product page for that product.

[0017] The advertisement distribution device 12 can post product advertisements on one or more channels selected by a merchant, for example. The merchant can select a channel on which to post an advertisement, taking into consideration the target users (i.e., purchasing demographic) of the product to be sold. Since the CVR and CTR may increase or decrease depending on the channel on which the product advertisement is posted, it is desirable to appropriately select the channel on which to post the advertisement. In addition to or instead of the merchant selecting a channel on which to post an advertisement, the advertisement distribution device 12 or the marketplace operation device 11 may select a channel on which to post an advertisement. Furthermore, the advertisement distribution device 12 or the marketplace operation device 11 may determine a channel on which to post an advertisement by machine learning, using user attributes (e.g., demographic information) and history information (e.g., information on visit activity and purchase activity for web pages, which will be described later) acquired from each user. Note that the advertisement distribution device 12 is not limited to a server device and may be realized by a mainframe or the like.

[0018] The user device 13 is operated by the user 15 to access the marketplace provided by the marketplace operating device 11 and receive various services. For example, the user 15 can operate the user device 13 to access the marketplace, browse product pages of various products offered in the marketplace, and purchase products on the product pages. In this embodiment, when the user 15 clicks on an advertisement page of a product distributed by the advertisement distribution device 12 and displayed on the user device 13 (including an operation of selecting an advertisement page displayed on the screen), the page displayed on the user device 13 moves from the advertisement page to a landing page corresponding to the advertisement, i.e., a product page of the product on the advertisement page. In other words, by clicking on any advertisement page, the user 15 can access a product page of the product on the clicked advertisement page in the marketplace provided by the marketplace operating device 11. In this embodiment, it is assumed that the landing page is directly accessed in response to a click on the advertisement page; however, the landing page may also be accessed in response to a step-by-step click on a web page accessed in response to a click on the advertisement page.

[0019] The user device 13 is a device such as a smartphone or tablet, and is configured to be able to communicate with the product list management device 10, the marketplace operation device 11, and the advertisement distribution device 12 via the network 14. The user device 13 has a display unit (display surface) such as a liquid crystal display, and the user 15 can view information such as web pages displayed on the display unit. The user 15 can also perform various operations using a GUI (Graphical User Interface) provided on the display unit. These operations include various operations on content such as images displayed on the screen, such as tapping, sliding, and scrolling using a finger or stylus. The user device 13 may also have a separate display unit.

[0020] The marketplace operation device 11 and the advertisement distribution device 12 are configured to collect various data in response to interactions with users 15 on web pages via user devices 13. The marketplace operation device 11 and the advertisement distribution device 12 each provide the collected various data to the product list management device 10. The data collected by the marketplace operation device 11 and the advertisement distribution device 12 will be described below. Note that while the data collected in response to interactions with users 15 will be described below, the marketplace operation device 11 and the advertisement distribution device 12 can collect data in response to interactions with multiple users, including user 15, via multiple user devices, and provide the data to the product list management device 10.

[0021] The marketplace operating device 11 collects, as visit activity, the behavior of the user 15 when browsing a product page. The visit activity is data representing the behavior of the user 15 on the product page of a product sold in the marketplace. The visit activity may include, for example, the number of accesses to the product page, the time spent on the product page, the page exit rate, and other behavioral patterns. The marketplace operating device 11 also collects, as purchase activity (also referred to as purchase history), purchases as an example of conversions to a product by the user 15. The purchase activity occurs when the user 15 purchases a product, and is registered as 1 (conversion occurred) if a purchase is made, or 0 (no conversion occurred) if no purchase is made. The purchase activity occurs, for example, when the user 15 adds a product to a cart on a product page, enters predetermined information, makes a payment, and purchases the product (i.e., performs a purchase action). The marketplace operating device 11 collects the visit activity and purchase activity as activity data and provides the collected activity data to the product list management device 10.

[0022] The advertisement distribution device 12 collects the number of times an advertisement page is displayed on a user device 13 as the number of impressions, and collects the number of clicks on the advertisement page by a user 15 as the number of clicks. The advertisement distribution device 12 can collect the number of impressions and the number of clicks as ad performance data for the advertisement page (hereinafter referred to as performance data). The performance data can be used to calculate a CTR. The CTR is a numerical value indicating the ratio of the number of clicks to the number of impressions for a web page. In this embodiment, the CTR corresponds to a numerical value indicating the ratio of the number of clicks to the number of impressions for a product advertisement page distributed by the advertisement distribution device 12. The advertisement distribution device 12 provides the collected performance data to the product list management device 10. The advertisement distribution device 12 may calculate the CTR and include it in the performance data provided to the product list management device 10.

[0023] Note that, as in this embodiment, when a landing page (i.e., a product page) is directly navigated to in response to a click on an advertising page, the number of accesses to the product page and the number of clicks on the advertising page have the same meaning and are the same number. That is, the number of accesses included in the visit activity included in the activity data and the number of clicks included in the performance data have the same meaning and are the same number. On the other hand, when a landing page is navigated to in response to gradual clicks on a web page navigated to in response to a click on an advertising page, the number of accesses to the product page and the number of clicks on the advertising page may not be the same number. That is, the number of accesses included in the visit activity included in the activity data and the number of clicks included in the performance data may not be the same number.

[0024] The product list management device 10 acquires from the merchants and stores a list of multiple (N (N is a natural number equal to or greater than 2)) products (hereinafter referred to as merchant product lists) that each merchant can sell in the marketplace operated by the marketplace operation device 11. The merchant product list may be catalog data of products offered by each merchant. Alternatively, the marketplace operation device 11 may manage the merchant product list and provide it to the product list management device 10. The products included in the merchant product list may be updated, for example, at regular intervals, and the update may be performed by the product list management device 10 or the marketplace operation device 11 at the instruction of the merchant.

[0025] The product list management device 10 acquires performance data from the advertisement distribution device 12 and activity data from the marketplace operation device 11. Based on the merchant product list, the performance data, and the activity data, the product list management device 10 calculates a product quality score for each of the N products included in the merchant product list. In this embodiment, the product quality score is a score indicating product quality (or product advertising quality) in terms of CVR and CTR, and corresponds to a score indicating the balance between CVR and CTR. That is, the product quality score corresponds to the cost indicating the balance between both conversion and advertising cost, and the higher the product quality score, the better the balance; specifically, the higher the CVR and CTR. The product list management device 10 calculates the product quality score using the actual conversion score (CV score) and CTR for the product. The CV score is a score indicating conversion, expressed as 1 (conversion) or 0 (no conversion). A product with a high product quality score can be expected to have a high CVR and a high CTR, i.e., a low advertising cost, and is beneficial to the merchant. In addition, since a high CTR is expected, this product can also be beneficial for operators of advertising platforms.

[0026] Then, the product list management device 10 selects M predetermined products (M < N, where N is a natural number) out of the N products included in the merchant product list as a predetermined number of products to be advertised. Then, a list of the selected M products is constructed (generated) as an optimized product list, and the optimized product list is provided to the advertisement distribution device 12. The product list management device 10 can perform such processes of calculating the product quality score, selecting products, constructing and providing the optimized product list for each merchant. The procedure for calculating the product quality score will be described later.

[0027] The advertisement distribution device 12 that has obtained the optimized product list posts and distributes advertisements for the M products included in the list on each channel. Which channel to post the advertisements for each of the M products may be determined by any method including machine learning. The advertisement distribution device 12 can perform such processes of obtaining the optimized product list and distributing the advertisements for each of the M products for each merchant.

[0028] A conceptual diagram of the overall processing by the product list management device 10, the marketplace operation device 11, and the advertisement distribution device 12 according to this embodiment is shown in FIG. 2. S20 represents the processing in the product list management device 10, S21 represents the processing in the advertisement distribution device 12, and S22 represents the processing in the marketplace operation device 11. FIG. 2 will be described in the order of the processing of S21, S22, and S20. Refer to FIG. 1 for the description of FIG. 2.

[0029] In this embodiment, it is assumed that a predetermined period (hereinafter referred to as the target period) for calculating a product quality score for each of the N products is set in advance. The advertisement distribution device 12 and the marketplace operation device 11 each collect performance data and activity data during the target period and provide them to the product list management device 10. The product list management device 10 then calculates a product quality score for each of the N products during the target period. For example, if the product quality score is calculated daily, the target period is a predetermined one-day period. The target period may be set in the product list management device 10 in advance, set by an operator, or set by a predetermined program, for example.

[0030] In S21, the advertisement distribution device 12 provides a plurality of channels in the advertisement platform. In the example of FIG. 2, a search site channel 211, an email channel 212, and a map site channel 213 are illustrated. The advertisement distribution device 12 can distribute advertisement pages by posting (i.e., uploading) product advertisements on at least one of the search site channel 211, the email channel 212, and the map site channel 213. The advertisement distribution device 12 collects the number of clicks and the number of impressions on the advertisement pages of each channel during a target period, generates performance data including the collected number of clicks and the number of impressions, and provides the performance data to the product list management device 10. In addition, the advertisement distribution device 12 can acquire an optimized list from the product list management device 10, and distribute advertisements for M (a plurality of) products included in the optimized list on at least one of the search site channel 211, the email channel 212, and the map site channel 213.

[0031] In S22, the marketplace operation device 11 collects visit activities and purchase activities during the target period on the product pages of the marketplace it provides, generates activity data including the collected visit activities and purchase activities, and provides the activity data to the product list management device 10. In this embodiment, the product pages correspond to landing pages that are displayed on the user device 13 in response to the user 15 clicking on the advertisement page provided by the advertisement distribution device 12.

[0032] In S20, the product list management device 10 acquires a merchant product list from the merchant. Note that the product list management device 10 may acquire the merchant product list from the merchant via another device. The product list management device 10 then calculates a product quality score for the target period using performance data acquired from the advertisement distribution device 12 and activity data acquired from the marketplace operation device 11. The procedure for calculating the product quality score will be described later. The product list management device 10 then uses the calculated product quality score to optimize the merchant product list, thereby constructing an optimized product list for the target period. Furthermore, the product list management device 10 provides the constructed optimized product list to the advertisement distribution device 12.

[0033] [Configuration of product list management device] FIG. 3 shows an example of the functional configuration of the product list management device 10 according to this embodiment. As an example of its functional configuration, the product list management device 10 has a performance data acquisition unit 301, an activity data acquisition unit 302, a score calculation unit 303, a list optimization unit 304, a product list storage unit 310, and a parameter storage unit 320. The product list storage unit 310 is configured to be able to store a merchant product list 311 and an optimized product list 312 for each merchant. FIG. 3 shows the merchant product list 311 and optimized product list 312 for a certain merchant. The parameter storage unit 320 is configured to be able to store various parameters used to calculate product quality scores.

[0034] The performance data acquisition unit 301 acquires performance data for a target period from the advertisement distribution device 12. The performance data includes the number of clicks and the number of impressions for the target period. The activity data acquisition unit 302 acquires activity data for the target period from the marketplace operation device 11. The activity data includes visit activity and purchase activity for the target period.

[0035] The score calculation unit 303 calculates a product quality score for each of N products for the target period using performance data and activity data acquired from multiple users, including user 15. In this embodiment, the score calculation unit 303 calculates the product quality score based on a bandit algorithm. A bandit algorithm is an algorithm classified as reinforcement learning in machine learning, and is an algorithm that aims to maximize rewards while balancing "exploitation" and "exploration." In this embodiment, the score calculation unit 303 calculates the product quality score based on an upper-confidence bound (UCB) algorithm, which is one of the bandit algorithms. In the UCB algorithm, the reward is calculated by adding an upper-limit estimated score indicating uncertainty (a value indicating "exploration") to a score based on actual data (a value indicating "exploitation"). In this embodiment, the upper-limit estimated score indicating uncertainty is a score obtained by weighting the estimated score with a weight indicating uncertainty (specifically, a weight value indicating the upper score limit).

[0036] The score calculation unit 303 calculates a predicted UCB (Upper Limit CV) for each product according to the UCB algorithm. cv (Hereinafter, PUCB cv ) and the predicted upper CTR (the upper limit of the predicted CTR)ctr (Hereinafter, PUCB ctr The predicted upper limit CV may be understood as the predicted upper limit CVR, which indicates the upper limit of the predicted CVR. Then, the score calculation unit 303 calculates the PUCB for each product. cv and P.U.C.B. ctr By multiplying this, the product quality score PUCB cvtr Calculate.

[0037] PUCB cv and P.U.C.B. ctr First, let us explain in more detail about PUCB. cv The score calculation unit 303 calculates the PUCB score for the product i according to the formula (1). cv PUCB cv (i) is calculated. Note that the index i and index j for a product used in this description represent an index that indicates one of N products included in a merchant product list provided by a given merchant.

number

[0038] CV(i) is the actual CV score for product i. The score calculation unit 303 can obtain CV(i) using the purchase activity included in the activity data for product i obtained from the marketplace operating device 11. CV(i) is 1 if at least one user out of multiple users purchases product i during the target period, and is 0 if none of the multiple users purchases product i. Ccv is a fixed value (hyperparameter) and is a parameter that controls machine learning for Mcv(i). N clk (i) is the number of clicks on the advertising page of product i, and Σ j N clk (j) is the total number of clicks on the advertisement pages of N items included in the merchant item list. The score calculation unit 303 calculates N from the number of clicks included in the performance data for item j (j=1 to N) acquired from the advertisement distribution device 12. clk (i) and Σ j N clk (j) can be calculated.

[0039] Mcv(i) is a CV score for product i estimated using a learning model for machine learning. In this embodiment, the score calculation unit 303 can calculate Mcv(i) using a trained CV(i) estimation model. The CV(i) estimation model may be a machine learning model based on a CatBoost (Category Boosting) binary classification model to estimate whether product i will acquire at least one conversion. The CV(i) estimation model is trained in advance using the number of clicks on past advertising pages for product i and purchase activity for product i, and parameters for configuring the CV(i) estimation model derived by training are stored in the parameter storage unit 320. In this embodiment, the product list management device 10 trains the CV(i) estimation model in advance using the number of clicks and purchase activity collected before calculating the initial product quality score, generates parameters for configuring the model, and stores them in the parameter storage unit 320. Note that training of the CV(i) estimation model may be performed by a device other than the product list management device 10.

[0040] Next, P.U.C.B. ctr The score calculation unit 303 calculates the PUCB score for the product i according to the formula (2). ctr PUCB ctr Calculate (i).

number

[0041] CTR(i) is the actual CTR for product i. The score calculation unit 303 can calculate CTR(i) using the number of impressions and the number of clicks included in the performance data for product i acquired from the advertisement distribution device 12. Cctr(i) is a fixed value (hyperparameter) and is a parameter that controls machine learning for Mctr(i). N imp (i) is the number of impressions for product i, and Σ j N imp (j) is the total number of impressions for N items included in the merchant item list. The score calculation unit 303 calculates N from the number of impressions included in the performance data for item j (j=1 to N) acquired from the advertisement distribution device 12. imp (i) and Σ j N imp (j) can be calculated.

[0042] Mctr(i) is the CTR for product i estimated using a learning model for machine learning. In this embodiment, the score calculation unit 303 can calculate Mctr(i) using a trained CTR(i) estimation model. The CTR(i) prediction model may be a machine learning model based on the CatBoost regression model. The CTR(i) estimation model is trained in advance using the past number of clicks and number of impressions for product i, and parameters for configuring the CTR(i) prediction model derived through training are stored in the parameter storage unit 320. In this embodiment, the product list management device 10 trains the CTR(i) estimation model in advance using the number of clicks and number of impressions collected before calculating the initial product quality score, generates parameters for configuring the model, and stores them in the parameter storage unit 320. Note that the training of the CTR(i) estimation model may be performed by a device other than the product list management device 10.

[0043] The score calculation unit 303 calculates PUCB(i) cv and PUCB(i) ctr After calculating the product quality score PUCB(i) of product i according to equation (3), cvtr That is, PUCB(i) cv and PUCB(i) ctr Product quality score PUCB(i) is calculated by multiplying cvtr Calculate.

number

[0044] The score calculation unit 303 calculates the product quality score PUCB(i) for all N products included in the merchant product list 311 during the target period. cvtr The score calculation unit 303 calculates the product quality score PUCB(i) cvtr may be calculated periodically, for example, for each target period. For example, if the target period is a predetermined one day, the score calculation unit 303 may calculate the product quality score PUCB(i) cvtr may be calculated and updated daily.

[0045] The list optimization unit 304 calculates the product quality scores PUCB(i) calculated for the N products included in the merchant product list 311. cvtr In this embodiment, the list optimization unit 304 selects a predetermined number M (N>N) of products to be advertised from the N products based on the product quality score PUCB(i) cvtr The list optimization unit 304 selects M products with higher scores by arranging them in descending order. Note that the selection of M products with higher scores is not limited to this, and for example, the list optimization unit 304 may select (Mm) products with higher scores (M>m) and m products with lower scores in order to verify the balance between conversion and advertising costs.

[0046] The list optimization unit 304 may select M products for each target period. That is, the list optimization unit 304 may select M products for each target period. cvtr Additionally or alternatively, the list optimizer 304 may select M products for each of a plurality of target periods. For example, the list optimizer 304 may select M products for each of a plurality of product quality scores PUCB(i) calculated for a plurality of target periods. cvtr Average the average PUCB(i) cvtr M items may be selected based on the above. Also, M does not have to be a fixed number, but may be a variable number.

[0047] The list optimization unit 304 selects M items from the N items included in the merchant item list 311 and constructs a list of the M items as an optimized item list 312. The list optimization unit 304 stores the optimized item list 312 in the item list storage unit 310 and calculates the item quality scores PUCB(i) for the N items. cvtr Each time the calculated value is used, the optimized product list 312 can be updated. The list optimization unit 304 provides the constructed optimized product list 312 to the advertisement distribution device 12.

[0048] [Hardware configuration of product list management device] Next, a description will be given of an example of the hardware configuration of the product list management device 10. Fig. 4 is a block diagram showing an example of the hardware configuration of the product list management device 10 according to this embodiment. The product list management device 10 according to this embodiment can be implemented on any single or multiple computers, mobile devices, or any other processing platform. 4, the product list management device 10 is shown implemented in a single computer, but the product list management device 10 according to this embodiment may be implemented in a computer system including multiple computers. The multiple computers may be connected to each other via a wired or wireless network so as to be able to communicate with each other.

[0049] 4, the product list management device 10 may include a CPU (Central Processing Unit) 401, a ROM (Read Only Memory) 402, a RAM (Random Access Memory) 403, a HDD (Hard Disk Drive) 404, an input unit 405, a display unit 406, a communication I / F (communication unit) (interface) 407, and a system bus 408. The product list management device 10 may also include an external memory. The CPU 401 controls the overall operation of the product list management device 10, and controls each component (402 to 407) via a system bus 408, which is a data transmission path.

[0050] The ROM 402 is a non-volatile memory that stores a control program and the like required for the CPU 401 to execute processing. The program includes instructions (code) for executing the processing according to the above-described embodiment. The program may be stored in a non-volatile memory such as the HDD 404 or an SSD (Solid State Drive), or in an external memory such as a removable storage medium (not shown). The RAM 403 is a volatile memory and functions as the main memory, work area, etc. of the CPU 401. That is, when executing a process, the CPU 401 loads necessary programs, etc. from the ROM 402 into the RAM 403 and executes the programs, etc. to realize various functional operations. The RAM 403 may include the product list storage unit 310 and the parameter storage unit 320 shown in FIG. 3.

[0051] The HDD 404 stores, for example, various data and information required when the CPU 401 performs processing using a program. The HDD 404 also stores, for example, various data and information obtained when the CPU 401 performs processing using a program. The input unit 405 is configured with a keyboard and a pointing device such as a mouse. The display unit 406 is configured by a monitor such as a liquid crystal display (LCD), etc. The display unit 406 may be configured in combination with the input unit 405 to function as a GUI (Graphical User Interface).

[0052] The communication I / F 407 is an interface that controls communication between the product list management device 10 and an external device. The communication I / F 407 provides an interface with a network and executes communication with the external device via the network. Various data, parameters, and the like are transmitted and received between the product list management device 10 and the external device via the communication I / F 407. In this embodiment, the communication I / F 407 may execute communication via a wired local area network (LAN) or a dedicated line that conforms to a communication standard such as Ethernet (registered trademark). However, the network that can be used in this embodiment is not limited to this and may be configured as a wireless network. This wireless network includes wireless personal area networks (PANs) such as Bluetooth (registered trademark), ZigBee (registered trademark), and UWB (Ultra Wide Band). It also includes wireless local area networks (LANs) such as Wi-Fi (Wireless Fidelity) (registered trademark) and wireless metropolitan area networks (MANs) such as WiMAX (registered trademark). It also includes wireless wide area networks (WANs) such as 4G and 5G. The network may be any network that connects devices to each other so that they can communicate with each other, and the communication standard, scale, and configuration are not limited to those described above.

[0053] At least some of the functions of each element of the product list management device 10 shown in Fig. 3 can be realized by the CPU 401 executing a program. However, at least some of the functions of each element of the product list management device 10 shown in Fig. 3 may be operated as dedicated hardware. In this case, the dedicated hardware operates under the control of the CPU 401.

[0054] The hardware configurations of the user device 13, the marketplace operating device 11, and the advertisement distribution device 12 may also be similar to the configuration shown in FIG.

[0055] [Product list management device processing flow] 5 shows a flowchart of processing executed by the product list management device 10 according to this embodiment. In S51, the performance data acquisition unit 301 acquires performance data for the target period from the advertisement distribution device 12, and the activity data acquisition unit 302 acquires activity data for the target period from the marketplace operation device 11.

[0056] In S52, the score calculation unit 303 calculates a product quality score for each of the N products in the target period based on the performance data and the activity data. The procedure for calculating the product quality score is as described above. In S53, the list optimization unit 304 selects M products to be advertised from the N products based on the calculated product quality score for each of the N products, and constructs an optimized product list 312. The list optimization unit 304 may select M products for each target period, or may select M products for multiple target periods. Then, in S54, the list optimization unit 304 provides the constructed optimized product list 312 to the advertisement distribution device 12. This allows the advertisement distribution device 12 to generate and distribute advertising pages for the M products included in the optimized product list 312. If the product list management device 10 and the advertisement distribution device 12 are configured as a single device, advertising pages for the M products included in the constructed optimized product list 312 can be distributed.

[0057] Through this process, the product list management device 10 can select products for advertising that can be expected to have a high probability of conversion and a high CTR. In other words, it becomes possible for merchants to select appropriate products that take into consideration the balance between conversion and advertising costs.

[0058] A conceptual diagram of products and product quality scores is shown in FIG. 6. In the example of FIG. 6, for products 60 and 61, product quality scores have been calculated. The product quality score of product 60 is 0.7, and the product quality score of product 61 is 0.4. Under such conditions, when the list optimization unit 304 selects either of the two products to construct an optimized product list, it can select product 60 which has a higher product quality score. Although the price of product 60 is lower than that of product 61, because the product quality score of product 60 is higher, product 60 can be expected to have a high conversion rate and a high CTR, and high profits can be expected for both the advertising distribution side (for example, the operator of the advertising platform) and the product providing side (for example, the merchant).

[0059] In this way, in order to select M (M < N) products that can be expected to have a high conversion rate and a high CTR from among N products as products to be advertised, the product list management apparatus 10 can achieve more beneficial advertising distribution for both the advertising distribution side and the product providing side than when selecting products to be advertised based only on CVR or CTR. Further, since the product list management apparatus 10 calculates the product quality score using the factual data collected by each of the advertising distribution side and the product providing side which are independent institutions, it can select a product that appropriately evaluates the advertising cost and the advertising effect without being advantageous to either side, and the reliability of this selection is enhanced. Also, the advertising distribution apparatus 12 reduces the processing load by distributing M advertising pages which are less than N instead of advertising pages for all N products.

[0060] In the above embodiment, the product list is optimized for each merchant. However, this is not limited to this. The product list may be optimized for each category, such as for each product category. For example, the product list management device 10 may calculate a product quality score for each of N products in a certain category according to the above procedure, and select N products to advertise based on the product quality scores. The product list may also be optimized for each merchant and for each category. For example, the product list management device 10 may calculate a product quality score for each of N products in a certain category that are available for sale by a certain merchant according to the above procedure, and select N products to advertise based on the product quality scores. In addition, although the above embodiment focuses on products sold in a marketplace operated by the marketplace operation device 11, this embodiment can be applied to any item that can be traded online, such as intangible services.

[0061] In the above embodiment, optimization may be performed using visit activity. For example, when a user moves to a landing page in response to a step-by-step click on a web page that has been moved in response to a click on an advertisement page, the PUCB (Pulse Response Block) shown in Equations (1) and (2) can be calculated. cv and P.U.C.B. ctr When calculating PUCB, the number of accesses to the product page included in the visit activity may be used instead of the number of clicks on the advertising page. In addition, the weights can be calculated by quantifying the time spent on the product page included in the visit activity, the page exit rate, and other behavioral patterns, and then the PUCB shown in equations (1) and (2) can be calculated. cv and P.U.C.B. ctr This may be reflected in the weights representing the uncertainty in each. For example, if the time spent on a product page is longer than a predetermined time and / or if the bounce rate is lower than a predetermined value, the product list management device 10 may determine the weights so that the weights representing the uncertainty are higher. This increases the second terms on the right-hand sides of equations (1) and (2), resulting in a higher product quality score.

[0062] In addition, in the above embodiment, an example of selecting a product to advertise on an online advertising page was described, but the advertising means for the selected product is not limited to an online advertising page, and may be, for example, paper media or other media.

[0063] In the above embodiment, the PUCB cv However, the calculation is not limited to the product page, as long as the purchasing activity for the product is acquired. For example, the purchasing activity for the product may be registered by the marketplace operating device 11 in response to the user 15 purchasing a product using voice or other information notified to the user device 13 in response to a click on an advertisement page.

[0064] Although specific embodiments have been described above, these embodiments are merely examples and are not intended to limit the scope of the present invention. The devices and methods described herein may be embodied in forms other than those described above. Furthermore, appropriate omissions, substitutions, and modifications may be made to the above-described embodiments without departing from the scope of the present invention. Such omissions, substitutions, and modifications are included within the scope of the claims and their equivalents, and belong to the technical scope of the present invention.

[0065] The disclosure of this embodiment includes the following configuration. [1] An information processing device having: a first data acquisition unit that acquires first data including the number of impressions and the number of clicks for an advertising page of each of a plurality of products during a predetermined period; a second data acquisition unit that acquires second data including purchase history for products advertised on advertising pages clicked during the predetermined period; a calculation unit that calculates a product quality score for each of the plurality of products during the predetermined period using the first data and the second data; and a selection unit that selects a predetermined number of products to be advertised from the plurality of products based on the product quality scores.

[0066] [2] The information processing device according to [1], wherein the selection unit selects the predetermined number of products from the plurality of products in descending order of the product quality score.

[0067] [3] The information processing device described in [1] or [2], wherein the calculation unit calculates a predicted upper limit CVR, which is an upper limit of a predicted CVR (Conversion Rate), and a predicted upper limit CTR, which is an upper limit of a predicted CTR (Click Through Rate), for each of the plurality of products using the first data and the second data, and calculates the product quality score by multiplying the predicted upper limit CVR and the predicted upper limit CTR.

[0068] [4] The information processing device according to any one of [1] to [3], wherein the selection unit selects the predetermined number of products from the plurality of products each time the product quality score for each of the plurality of products is calculated.

[0069] [5] An information processing device described in any of [1] to [3], wherein the selection unit selects the predetermined number of products from the plurality of products each time a plurality of product quality scores for each of the plurality of products are calculated.

[0070] [6] The information processing device according to any one of [1] to [5], further comprising an advertisement distribution unit that distributes advertisements for the predetermined number of products. [Explanation of symbols]

[0071] 10: Product list management device, 11: Marketplace operation device, 12: Advertisement distribution device, 13: User device, 14: Network, 15: User, 301: Performance data acquisition unit, 302: Activity data acquisition unit, 303: Score calculation unit, 304: List optimization unit, 310: Product list storage unit, 311: Merchant product list, 312: Optimized product list, 320: Parameter storage unit

Claims

1. a first data acquisition unit that acquires first data including the number of impressions and the number of clicks for each advertising page of a plurality of (N) products during a predetermined period; a second data acquisition unit that acquires second data including a purchase history of a product advertised on the clicked advertising page during the predetermined period; a calculation unit that calculates a product quality score that indicates a balance between a conversion rate (CVR) and a click through rate (CTR) for each of the plurality of products during the predetermined period, using the first data and the second data; a selection unit that selects a predetermined number of products to be advertised from the plurality of products based on the product quality scores; and the calculation unit calculates, for each of the plurality of products, a predicted upper limit CVR, which is an upper limit of a predicted CVR, and a predicted upper limit CTR, which is an upper limit of a predicted CTR, using the first data and the second data, and calculates the product quality score by multiplying the predicted upper limit CVR and the predicted upper limit CTR; the predicted upper limit CVR is a value based on a CV score indicating whether or not each of the plurality of products has actually been purchased, which is obtained using the second data, and a CV score for each of the plurality of products estimated using a first learning model for machine learning; the predicted upper limit CTR is a value based on an actual CTR for each of the plurality of products acquired using the first data and a CTR for each of the plurality of products estimated using a second learning model for machine learning; The predicted upper limit CVR (PUCB cv (i)) for each of the plurality of products i is expressed by equation (1), where CV(i) is a CV score representing whether or not product i of each of the plurality of products has actually been purchased, Ccv is a fixed parameter, Mcv(i) is a CV score for product i of each of the plurality of products estimated using the first learning model, √Σ j N clk (j) / 1+N clk (j) is a weight representing uncertainty, N clk (i) is the number of clicks on the advertising page of product i, and Σ j N clk (j) is the total number of clicks on the advertising pages of the plurality of products j (j = 1 to N), The predicted upper limit CTR (PUCB ctr (i)) for each product i of the plurality of products is expressed by equation (2): Here, CTR(i) is the actual CTR for product i of each of the multiple products, Cctr is a fixed parameter, Mctr(i) is the CTR for product i of each of the multiple products estimated using the second learning model, √Σ j N imp (j) / 1 + N imp (j) is a weight representing uncertainty, N imp (i) is the number of impressions for product i, and Σ j N imp (j) is the total number of impressions for the multiple products j (j = 1 to N).

2. the selection unit selects the predetermined number of products from the plurality of products in descending order of the product quality score. The information processing device according to claim 1 .

3. the selection unit selects the predetermined number of products from the plurality of products each time the product quality score for each of the plurality of products is calculated. The information processing device according to claim 1 .

4. the selection unit selects the predetermined number of products from the plurality of products each time the plurality of product quality scores for each of the plurality of products are calculated. The information processing device according to claim 1 .

5. further comprising an advertisement distribution unit that distributes advertisements for the predetermined number of products; The information processing device according to claim 1 .

6. An information processing method executed by an information processing device, acquiring first data including the number of impressions and the number of clicks for an advertisement page of each of a plurality of products during a predetermined period; acquiring second data including a purchase history of products advertised on the advertising page clicked during the predetermined period; calculating a product quality score representing a balance between a conversion rate (CVR) and a click through rate (CTR) for each of the plurality of products during the predetermined period using the first data and the second data; selecting a predetermined number of products to be advertised from the plurality of products based on the product quality scores; Including, the calculating step includes calculating, for each of the plurality of products, a predicted upper limit CVR, which is an upper limit of a predicted CVR, and a predicted upper limit CTR, which is an upper limit of a predicted CTR, using the first data and the second data, and calculating the product quality score by multiplying the predicted upper limit CVR and the predicted upper limit CTR; the predicted upper limit CVR is a value based on a CV score indicating whether or not each of the plurality of products has actually been purchased, which is obtained using the second data, and a CV score for each of the plurality of products estimated using a first learning model for machine learning; the predicted upper limit CTR is a value based on an actual CTR for each of the plurality of products acquired using the first data and a CTR for each of the plurality of products estimated using a second learning model for machine learning; The predicted upper limit CVR (PUCB cv (i)) for each of the plurality of products i is expressed by equation (1), where CV(i) is a CV score representing whether or not product i of each of the plurality of products has actually been purchased, Ccv is a fixed parameter, Mcv(i) is a CV score for product i of each of the plurality of products estimated using the first learning model, √Σ j N clk (j) / 1+N clk (j) is a weight representing uncertainty, N clk (i) is the number of clicks on the advertising page of product i, and Σ j N clk (j) is the total number of clicks on the advertising pages of the plurality of products j (j = 1 to N), The predicted upper limit CTR (PUCB ctr (i)) for each product i of the plurality of products is expressed by equation (2): Here, CTR(i) is the actual CTR for product i of each of the multiple products, Cctr is a fixed parameter, Mctr(i) is the CTR for product i of each of the multiple products estimated using the second learning model, √Σ j N imp (j) / 1 + N imp (j) is a weight representing uncertainty, N imp (i) is the number of impressions for product i, and Σ j N imp (j) is the total number of impressions for the multiple products j (j = 1 to N).

7. An information processing program for causing a computer to execute information processing, the program including: a first data acquisition process for acquiring first data including the number of impressions and the number of clicks for each advertising page of a plurality of products during a predetermined period; a second data acquisition process for acquiring second data including a purchase history of a product advertised on the clicked advertising page during the predetermined period; a calculation process for calculating a product quality score representing a balance between a conversion rate (CVR) and a click through rate (CTR) for each of the plurality of products during the predetermined period, using the first data and the second data; a selection process of selecting a predetermined number of products to be advertised from the plurality of products based on the product quality score, the calculation process includes calculating, for each of the plurality of products, a predicted upper limit CVR that is an upper limit of a predicted CVR and a predicted upper limit CTR that is an upper limit of a predicted CTR, using the first data and the second data, and calculating the product quality score by multiplying the predicted upper limit CVR and the predicted upper limit CTR; the predicted upper limit CVR is a value based on a CV score indicating whether or not each of the plurality of products has actually been purchased, which is obtained using the second data, and a CV score for each of the plurality of products estimated using a first learning model for machine learning; the predicted upper limit CTR is a value based on an actual CTR for each of the plurality of products acquired using the first data and a CTR for each of the plurality of products estimated using a second learning model for machine learning; The predicted upper limit CVR (PUCB cv (i)) for each of the plurality of products i is expressed by equation (1), where CV(i) is a CV score representing whether or not product i of each of the plurality of products has actually been purchased, Ccv is a fixed parameter, Mcv(i) is a CV score for product i of each of the plurality of products estimated using the first learning model, √Σ j N clk (j) / 1+N clk (j) is a weight representing uncertainty, N clk (i) is the number of clicks on the advertising page of product i, and Σ j N clk (j) is the total number of clicks on the advertising pages of the plurality of products j (j = 1 to N), The predicted upper limit CTR (PUCB ctr (i)) for each product i of the plurality of products is expressed by equation (2): where CTR(i) is the actual CTR for product i of each of the multiple products, Cctr is a fixed parameter, Mctr(i) is the CTR for product i of each of the multiple products estimated using the second learning model, √Σ j N imp (j) / 1+N imp (j) is a weight representing uncertainty, N imp (i) is the number of impressions for product i, and Σ j N imp (j) is the total number of impressions for the multiple products j (j = 1 to N). Information processing program.

Images