JP6976207B2 - Information processing equipment, information processing methods, and programs - Google Patents

Description

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

When a consumer purchases a product or service sold on a shopping site or the like, a technique is known that recommends the consumer to purchase other products or services together (for example,). See Patent Document 1).

Japanese Unexamined Patent Publication No. 2018-10559

However, in the conventional technology, a product or service recommended in anticipation of a predetermined action such as a purchase cannot sufficiently motivate consumers to take a predetermined action. There was a case.

The present invention has been made in view of the above problems, and provides an information processing device, an information processing method, and a program capable of more accurately recommending a product or service in which a consumer is likely to take a predetermined action. The purpose is to do.

In one aspect of the present invention, the behavior history of a user who has reached a predetermined behavior for a product or service sold via a network is summarized for each processing reference period based on a series of behaviors of the user. The first generation unit that generates one or more first data and the goods or services included in each of the one or more first data generated by the first generation unit are grouped into one for each category. Of the goods or services included in each of the second generation unit that generates one or more second data and the one or more second data generated by the second generation unit, the same processing reference period. It is an information processing apparatus including a third generation unit that generates one or more third data in which a plurality of the products or services that the user has reached a predetermined action are put together into one.

According to one aspect of the present invention, it is possible to more accurately recommend a product or service in which a consumer is likely to take a predetermined action.

It is a figure which shows an example of the information processing system 1 including the information processing apparatus 100 in embodiment. It is a figure which shows an example of the web page provided by the service providing apparatus 20. It is a figure which shows the other example of the web page provided by the service providing apparatus 20. It is a figure which shows an example of the structure of the information processing apparatus 100 in an embodiment. It is a flowchart which shows the flow of a series of processing by a delivery management unit 102 and an optimization processing unit 104 in an embodiment. It is a figure for demonstrating the component of the database generation processing part 110 in Embodiment, and the processing content of each component. It is a figure which shows an example of the action history data. It is a figure which shows an example of the 1st aggregated log data. It is a figure which shows an example of the 2nd aggregated log data. It is a figure for demonstrating the method of generating the recommended product candidate list data. It is a flowchart which shows the flow | flow of the series of processing by the database update part 106 and the database generation processing part 110 in embodiment. It is a figure for demonstrating the determination method of the similar product with respect to the recommended product. It is a figure which shows an example of the web page which the advertisement content was delivered. It is a figure which shows an example of the hardware composition of the information processing apparatus 100 of an embodiment.

Hereinafter, an information processing apparatus, an information processing method, and a program to which the present invention is applied will be described with reference to the drawings.

[Overview]
The information processing device is realized by one or more processors. The information processing device is a plurality of user's action histories that have reached a predetermined action for a product or service sold via a network, and are collected into one for each processing standard period based on a series of user actions. The first data of is generated. A product or service sold over a network is, for example, a product or service sold on a website or the like. The information processing device generates a plurality of second data in which the products or services included in each of the generated plurality of first data are grouped into one for each category, and each of the plurality of second data. Among the products or services included in the above, a third data is generated in which a plurality of products or services in which the user has reached a predetermined action within the same processing standard period are grouped together.

In the third data, a plurality of products or services in which an unspecified number of users have reached a predetermined action within the same predetermined period are grouped by category. Therefore, for example, a target user who is a consumer is a product. Or, when the service is browsed, in the third data, the other goods or services associated with the goods or services browsed by the target user are in the same category as the goods or services browsed by an unspecified number of users. In addition, it is treated as a product or service that has reached a predetermined action within the same predetermined period. A product or service having such a property can be a product or service that is likely to cause a predetermined action even for the target user. Therefore, the information processing device extracts from the generated third data other products or services associated with the products or services viewed by the target user, so that the target user can easily take a predetermined action. The service can be recommended more accurately by the target user. Hereinafter, the goods or services will be collectively referred to as “goods” and described.

[overall structure]
FIG. 1 is a diagram showing an example of an information processing system 1 including an information processing apparatus 100 according to an embodiment. The information processing system 1 in the embodiment includes, for example, one or more terminal devices 10, a service providing device 20, and an information processing device 100. These devices are connected via the network NW. Further, a part of these devices may be included in another device as a virtual device. For example, a part or all of the functions of the service providing device 20 are realized by the functions of the information processing device 100. The information processing device 100 may be a virtual machine, or conversely, a virtual machine in which a part or all of the functions of the information processing device 100 may be realized by the functions of the service providing device 20.

Each device shown in FIG. 1 transmits and receives various information via the network NW. The network NW includes, for example, the Internet, a WAN (Wide Area Network), a LAN (Local Area Network), a provider terminal, a wireless communication network, a wireless base station, a dedicated line, and the like. It should be noted that not all combinations of the devices shown in FIG. 1 need to be able to communicate with each other, and the network NW may include a local network in part.

The terminal device 10 is a terminal device including an input device, a display device, a communication device, a storage device, and an arithmetic unit, such as a mobile phone such as a smartphone, a tablet terminal, and various personal computers. The communication device includes a network card such as a NIC (Network Interface Card), a wireless communication module, and the like. In the terminal device 10, a UA (User Agent) such as a web browser or an application program is activated, and a request according to the content input by the user is transmitted to the service providing device 20. Further, the terminal device 10 in which the UA is activated causes the display device to display various images based on the information acquired from the service providing device 20.

The service providing device 20 is, for example, a web server that provides a web page to the terminal device 10 in response to a request from a web browser activated as a UA. The web page may be, for example, a web page constituting a website such as a shopping site, an auction site, or a free market site that sells products on the Internet (hereinafter, collectively referred to as a "sales site"). Further, the service providing device 20 may provide the terminal device 10 with a web page that provides various services such as a search site, an SNS (Social Networking Service), and a mail service. Further, the service providing device 20 is an application server that provides the same service as various websites such as a sales site by providing contents to the terminal device 10 in response to a request from an application started as a UA. May be good.

FIG. 2 is a diagram showing an example of a web page provided by the service providing device 20. In the illustrated example, one web page (referred to as a cart list page) of a sales site listing the products inserted into the shopping card as the products that the user wants to purchase is schematically shown. On such a cart list page, as shown in the area of R1 in the figure, products recommended to be purchased together with the products inserted in the shopping card are posted. The "combined purchase" may be, for example, confirming (simultaneously purchasing) the orders of a plurality of products placed in the shopping cart at the same timing, or each of the plurality of products placed in the shopping cart. The order may be completed within a certain fixed period at any time.

FIG. 3 is a diagram showing another example of a web page provided by the service providing device 20. In the illustrated example, one web page (hereinafter referred to as a product detail page) of a sales site on which detailed product information is posted is schematically shown. Such a product detail page includes contents such as an image of a product (R2 in the figure), a sentence indicating the title of the product (R3 in the figure), and a sentence indicating a description or outline of the product (R4 in the figure). In addition, products that are recommended for combined purchase are listed (R5 in the figure).

When the user (consumer) causes a predetermined action expected by the seller on the sales site provided by the service providing device 20, the information processing device 100 determines a product that the user is likely to reach the predetermined action. .. A predetermined action is an action that is a condition for establishing a conversion. For example, a product listed in a list format on a sales site is selected by a click operation or a touch operation, and the screen transitions to the product detail page of that product ( That is, browsing the product), putting the product in the shopping cart, listing the product on the cart list page, purchasing the product put in the shopping cart, and so on.

For example, in the information processing apparatus 100, when a user accesses a product detail page included in a sales site and browses a product, the user reaches a predetermined action for a product that is easy to be browsed at the same time as the product browsed by the user. Decide as an easy product.

Further, for example, when the information processing apparatus 100 accesses the product detail page included in the sales site and browses the product, the information processing device 100 considers that the product browsed by the user has been put into the shopping cart (assumed). ), A product that is likely to be put into the shopping cart at the same time as the product may be determined as a product that is likely to lead to a predetermined action by the user, or when it is considered that the product viewed by the user has been purchased. A product that is easy to be purchased at the same time as the product may be determined as a product that is easy for the user to take a predetermined action.

Further, for example, in the information processing apparatus 100, instead of the user accessing the product detail page included in the sales site and browsing the product, or in addition, the user puts the product into the shopping cart from the product detail page. In this case, a product that is likely to be put into the shopping cart at the same time as a product that the user has put into the shopping cart may be determined as a product that is easy for the user to take a predetermined action, or a product that the user has put into the shopping cart may be determined. When it is considered that the product has been purchased, a product that is likely to be purchased at the same time as the product may be determined as a product that is likely to lead to a predetermined action by the user.

Further, for example, when the user purchases a product put into a shopping cart, the information processing apparatus 100 may determine a product that is easily purchased at the same time as the product as a product that the user can easily reach a predetermined action. ..

[Information processing device configuration]
FIG. 4 is a diagram showing an example of the configuration of the information processing apparatus 100 in the embodiment. As shown in the figure, the information processing apparatus 100 includes, for example, a distribution management unit 102, an optimization processing unit 104, a database update unit 106, a database generation processing unit 110, a distribution setting database DB1, an inventory database DB2, and the like. It includes a product database DB 3, a first recommended product database DB 4, an action history database DB 5, and a second recommended product database DB 6. The distribution management unit 102 is an example of the “acquisition unit”, and the combination of the distribution management unit 102 and the optimization processing unit 104 is an example of the “recommendation processing unit”.

The distribution management unit 102, the optimization processing unit 104, the database update unit 106, and the database generation processing unit 110 execute a program stored in a storage device in which a processor such as a CPU (Central Processing Unit) realizes various databases. It is realized by doing. Further, a part or all of the distribution management unit 102, the optimization processing unit 104, the database update unit 106, and the database generation processing unit 110 are LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field). -It may be realized by hardware (circuitry) such as Programmable Gate Array) or GPU (Graphics Processing Unit), or it may be realized by the cooperation of software and hardware.

The distribution setting database DB1, inventory database DB2, product database DB3, first recommended product database DB4, action history database DB5, and second recommended product database DB6 are, for example, HDD (Hard Disc Drive), flash memory, EEPROM (Electrically Erasable). It is realized by storage devices such as Programmable Read Only Memory), ROM (Read Only Memory), and RAM (Random Access Memory). Various programs such as firmware and application programs are stored in these storage devices.

In the distribution setting database DB1, for example, products that are recommended to be purchased at the same time (combined purchase), products that are recommended to be added to the shopping cart at the same time, products that are recommended to be viewed at the same time, etc. are posted. A program (algorithm) for deciding what kind of product to be listed is stored in each frame. Hereinafter, products that are recommended to be purchased at the same time, products that are recommended to be added to the shopping cart at the same time, and products that are recommended to be viewed at the same time are collectively referred to as "recommended products".

The inventory database DB 2 includes data in which the number of products in stock handled by each store opened on the sales site is associated with the identification information (hereinafter referred to as product ID) of each product.

The product database DB 3 includes data in which product data related to products such as product titles, prices, and product categories of products handled at each store opened on the sales site are associated with product IDs.

The first recommended product database DB 4 and the second recommended product database DB 6 include product IDs of candidates for recommended products to be posted in the product posting frame.

The action history database DB 5 includes action history data in which the action history of the user is associated with a date and time (time stamp). For example, the action history database DB 5 may be updated on an hourly, daily, or monthly basis.

Upon receiving a predetermined request from the terminal device 10 that displays the product detail page or the cart list page, the distribution management unit 102 determines what kind of recommended product is to be posted in the product posting frame on each page. The predetermined request is, for example, information requesting the product data of the recommended product to be posted in the product posting frame, and the product ID of the product currently being browsed by the user in order to extract the product ID of the recommended product. The product ID of the product put into the shopping cart by the user and the product ID of the product purchased by the user are included.

The optimization processing unit 104 extracts product data of recommended products having properties determined by the distribution management unit 102 from various databases, and determines the placement order of recommended products in the product placement frame.

The database update unit 106 repeats the process of updating the first recommended product database DB 4 using the data included in the second recommended product database DB 6 in the first predetermined cycle (for example, once a day).

The database generation processing unit 110 generates a new recommended product list using the action history database DB 5, and adds the generated recommended product list to the second recommended product database DB 6, so that the second recommended product database DB 6 To update. This process is repeated in a second predetermined cycle (for example, once a day). The second predetermined cycle may be the same period as the first predetermined cycle, or may be a shorter period than the first predetermined cycle.

[Online processing flow]
Hereinafter, the flow of a series of processes (online processing or on-demand processing) by the distribution management unit 102 and the optimization processing unit 104 in the embodiment will be described according to a flowchart. FIG. 5 is a flowchart showing a flow of a series of processes by the distribution management unit 102 and the optimization processing unit 104 in the embodiment.

First, the distribution management unit 102 waits until a predetermined request is received from the terminal device 10 that displays the product detail page or the cart list page (S100), and when the predetermined request is received from the terminal device 10, the distribution setting database DB1 For each product placement frame, the nature of the recommended product to be posted in that frame is determined (S102).

For example, the distribution management unit 102 refers to the program stored in the distribution setting database DB1, and for the product placement frame (area R5 in FIG. 3) on the product detail page, the product (viewed by the user) posted on that page. When a product) is considered to have been purchased, a product with the first property that is easy to purchase at the same time as the product is posted, and the product placement frame (area R1 in FIG. 2) on the cart list page is displayed. A page that is displayed when an order for a product is confirmed by posting a product with a second property that is easy to purchase at the same time as the product when it is considered that the product put in the shopping cart has been purchased. Regarding the product posting frame (not shown), it is decided to post a product having a third property that is easy to be purchased at the same time as the product purchased by the user. Hereinafter, the product having the first property will be referred to as a view2conv product, and the product having the second property or the third property will be referred to as a conv2conv product.

Next, the optimization processing unit 104 uses the product ID (hereinafter referred to as the target product ID) included in the predetermined request received by the distribution management unit 102 as a key, and the property of the product determined for each posting frame. The product ID of the recommended product that matches the above is extracted from the first recommended product database DB4 (S104).

For example, when it is decided that the view2conv product is posted on the product posting frame on the product detail page, the optimization processing unit 104 determines the product IDs of the plurality of recommended products associated with the key product IDs. Among them, the product ID of the recommended product having the first property is extracted. Further, for example, when it is decided that the conv2conv product is to be posted on the product placement frame on the cart list page, the optimization processing unit 104 may use the optimization processing unit 104 to display the product of a plurality of recommended products associated with the key product ID. Among the IDs, the product ID of the recommended product having the second property or the third property is extracted.

Next, the optimization processing unit 104 extracts the product data of the recommended product from the inventory database DB 2 and the product database DB 3 by using the product ID of the extracted recommended product (S106). For example, the optimization processing unit 104 extracts the inventory quantity of the product associated with the key product ID from the inventory database DB2, and when the inventory quantity is 1 or more (that is, when the inventory exists). Further, product data including the title and price of the product associated with the key product ID is extracted from the product database DB3.

Next, the optimization processing unit 104 determines the placement order within the product placement frame based on the weight of each recommended product indicated by the extracted product ID (S108). The weight of the recommended product will be described later.

Next, the distribution management unit 102 transmits the product data of the recommended product whose placement order within the product placement frame is determined by the optimization processing unit 104 to the terminal device 10 as a response to a predetermined request together with the placement order (. S110). As a result, for example, from the viewpoint of the user who uses the terminal device 10, the recommended products having the highest ranking are arranged in order from the left side to the right side of the product placement frame or from the upper side to the lower side of the product placement frame. Will be posted.

[Configuration and processing content of database generation processing unit]
FIG. 6 is a diagram for explaining the components of the database generation processing unit 110 in the embodiment and the processing contents of each component. The database generation processing unit 110 in the embodiment includes, for example, a first data generation unit 112, a specific history extraction unit 114, and a second data generation unit 116. The first data generation unit 112 is an example of the "first generation unit", and the second data generation unit 116 is an example of the "second generation unit" and the "third generation unit".

The first data generation unit 112 acquires action history data from the action history database DB 5, collects the acquired action history data for each user ID, and further collects one action history data for each user ID for each session. By summarizing the data, the behavior history data for each user ID and each session are reorganized into one (hereinafter referred to as the first aggregate log data). The first aggregated log data is an example of "first data".

A session is, for example, a period from the time when the same user browses a certain product to the time when the browsed product is purchased. Further, if the product is not purchased within the third predetermined period (for example, one week) after the user browses the product, the session has a third predetermined period after the user browses the product. It may be a period up to (time-out period). A session is an example of a "processing reference period".

FIG. 7 is a diagram showing an example of behavior history data. As shown in the illustrated example, the action history data is a list of one record in which the history of actions taken by the user and the date and time when the action was taken are associated with the user ID that identifies the user. It is data. The behavior history includes, for example, a history of what kind of product was browsed on which sales site, what kind of product was put into the shopping cart, or what kind of product was purchased.

FIG. 8 is a diagram showing an example of the first aggregated log data. As shown in the illustrated example, in the first aggregate log data, data associated with the history of actions taken during the session period is collected for each user ID for each session ID that identifies each session. It is data. For example, a user browses product X on one of the sales sites, "○○! Shopping site", browses a plurality of products such as product Y and product Z in quick succession, and then purchases product X. In this case, the browsing history of a plurality of products is associated with one session ID, as in the record in which the session ID included in the first aggregate log data is "SESH_1". For example, in the first aggregated log data, for a certain user A, the product of the category "1" is grouped in the session whose session ID is "SESH_1", and the session whose session ID is "SESH_1" is the session. , Products with a category of "2" are grouped together, and products with a category of "3" are grouped together in a session with a session ID of "SESH_3".

The specific history extraction unit 114 extracts specific action history data from each of the user ID and the plurality of first aggregated log data generated for each session by the first data generation unit 112. The specific behavior history data is, for example, behavior history data indicating that the browsed product has been purchased. As a result, for example, when the behavior history data of a certain user is divided into sessions 1 to k, the purchase of the product browsed from the first aggregated log data of a plurality of sessions from 1 to k. The behavior history that did not reach is removed. As a result, only the action history leading to the purchase after browsing the product is left in the first aggregated log data. Further, the specific behavior history data may be, for example, behavior history data indicating that the browsed product has been put into the shopping cart. In this case, if the behavior history data of a certain user is divided into sessions 1 to k, the product is browsed and put into the shopping cart from the first aggregated log data of multiple sessions 1 to k. The action history that you did not do is removed. As a result, in the first aggregated log data, the behavior history put into the shopping cart after browsing the product or the behavior history leading to the purchase after browsing the product is left.

The second data generation unit 116 regroups a plurality of first aggregated log data including only specific action history data, that is, a plurality of first aggregated log data including the same product ID into one for each product category. , From the first aggregated log data, the second aggregated log data that summarizes the action history data for each product category is generated. The second aggregated log data is an example of "second data".

FIG. 9 is a diagram showing an example of the second aggregated log data. As shown in the illustrated example, the second aggregated log data is the first aggregated log data collected for each session such as "SESH_1", "SESH_2", and "SESH_3" with respect to the behavior history data of one user. Furthermore, the data collected for each product category is used as one unit. For example, the second aggregated log data may be data in which an action history is associated with a product ID that identifies each product for each category of the product.

Then, the second data generation unit 116 is data in which the second aggregated log data divided into the same session is combined among a plurality of second aggregated log data divided into product category units (hereinafter, recommended product candidate list). Data) is generated and stored in the second recommended product database DB6. The recommended product candidate list data is an example of "third data".

FIG. 10 is a diagram for explaining a method of generating recommended product candidate list data. For example, when focusing on a certain session i whose session ID is "SESH_i", the second aggregated log data in which the product category is classified into "shoes" and the product category are classified into "shirts" in the session i. It is assumed that the second aggregated log data is included. In this case, the same user has purchased a plurality of products in different categories during the same session.

Therefore, the second data generation unit 116 combines different categories of "shoes" and "shirts" into one, and extracts products included in the combined categories as candidates for recommended products. In other words, the second data generation unit 116 regards the products included in each of the plurality of different categories as products that are easily purchased together (or products that are easily put into the shopping cart at the same time). The products included in the category of are extracted as candidates for recommended products. Further, since the second aggregated log data having the session ID “SESH_k” is a session different from the second aggregated log data having the session ID “SESH_i”, the categories are not integrated into one in these data. In the illustrated example, the second aggregated log data includes other action histories such as "browsing" in addition to "purchase", but as described above, the action history of "purchase" is performed by the specific history extraction unit 114. Since only (specific action history) is extracted, the record associated with the action history such as "browsing" does not have to be included in the second aggregated log data in advance.

Then, the second data generation unit 116 generates the recommended product candidate list data in which the product IDs of the extracted products are associated with each other. In the illustrated example, each product (purchased product) whose product ID is "ITEM_A" to "ITEM_E" is classified into the product category of "shoes", and the product ID is "ITEM_O" to "ITEM_S". Each product (purchased product) is classified into the "shirt" product category. Since the behavior history data of purchasing each of these products is classified into the same session, in the recommended product candidate list data, multiple products with product IDs "ITEM_A" to "ITEM_E" and product IDs "ITEM_O" are used. ~ A plurality of "ITEM_S" products are associated with each other beyond the boundaries of categories as products that are easily purchased together or that are easily put into a shopping cart at the same time.

[Batch processing flow]
Hereinafter, the flow of a series of processes (batch processes) by the database update unit 106 and the database generation process unit 110 in the embodiment will be described according to a flowchart. FIG. 11 is a flowchart showing a flow of a series of processes by the database update unit 106 and the database generation processing unit 110 in the embodiment. As an example, the processing of this flowchart will be described assuming that the first predetermined cycle and the second predetermined cycle are the same period. When the first predetermined cycle and the second predetermined cycle are different periods from each other, the series of processes of S200 to S212 described later and the process of S214 may be performed independently (in parallel) with each other. ..

First, the first data generation unit 112 acquires the action history data from the action history database DB 5 (S200).

Next, the first data generation unit 112 collects the acquired action history data for each user ID, and further collects the action history data collected for each user ID into one for each session, so that the first aggregate log Generate data (S202).

Next, the specific history extraction unit 114 extracts specific action history data from each of the plurality of first aggregated log data generated by the first data generation unit 112, and removes other action history data. (S204).

Next, the second data generation unit 116 generates the second aggregated log data from the plurality of first aggregated log data including only the specific action history data (S206).

Next, the second data generation unit 116 combines the second aggregated log data divided into the same sessions to generate recommended product candidate list data (S208).

Next, the second data generation unit 116 updates the second recommended product database DB 6 by adding the product ID of the recommended product included in the recommended product candidate list data to the second recommended product database DB 6 (S210).

Next, the database update unit 106 updates the first recommended product database DB 4 by adding the product ID of the recommended product included in the second recommended product database DB 6 to the first recommended product database DB 4 (S212). This ends the processing of the flowchart.

[How to determine the ranking based on the weight of recommended products]
Hereinafter, a method of determining the ranking based on the weight of the recommended product by the optimization processing unit 104 described above will be described. For example, the optimization processing unit 104 extracts the product ID of the recommended product associated with the target product ID included in the predetermined request from the first recommended product database DB 4. Since the recommended product candidate list data is indirectly added to the first recommended product database DB 4, the product ID of the recommended product associated with the target product ID is used as the target product ID in the recommended product candidate list data. It becomes the associated product ID.

The optimization processing unit 104 may increase the weight of the recommended product as the product ID of the recommended product purchased more frequently during the same session among the extracted plurality of product IDs, or during the same session. The weight of the recommended product may be increased as the product ID of the recommended product is put into the shopping cart more frequently. Then, the optimization processing unit 104 raises the ranking of the determined recommended product with a larger weight, and lowers the ranking of the determined recommended product with a smaller weight. As a result, the higher the weight and the higher the ranking of the product, the more preferentially it will be placed in the product listing frame.

Further, the optimization processing unit 104 may be a recommended product in which a product similar to the recommended product in which the product ID is associated with the target product ID is posted in the product posting frame in the first recommended product database DB4. When the products are "similar" to each other, it means that the vectors of the products are close to each other in a certain vector space when the products to be compared are vectorized.

For example, the optimization processing unit 104 extracts text-based content representing a product title or product outline from a product detail page on which a recommended product in which a product ID is associated with a target product ID is posted, and uses the extracted content as the content. A multidimensional vector having statistics such as the frequency of appearance of each included word as each element may be generated as a vector of recommended products.

Further, the optimization processing unit 104 uses a method called distributed expression or word embedding expression (Word Embedding (s)) such as word2vec or GloVe, and uses a method called Word Embedding (s) to vector a recommended product based on each word included in the extracted content. May be generated. A distributed expression or word embedding expression is a method of vectorizing a word or phrase based on the co-occurrence of the word or phrase and the word or phrase that appears before or after the word or phrase, and is prepared in advance, for example. Based on a corpus containing multiple words or phrases, the appearance probability that appears before or after the word or phrase contained in the target content is calculated, and a multidimensional vector with the appearance probability as the element value is generated. be.

The optimization processing unit 104 derives the similarity (for example, cosine similarity) between each of the vectors of a plurality of similar products prepared in advance and the vector of the recommended product in which the product ID is associated with the target product ID. .. A similar product is a product that can be similar to a recommended product, and is, for example, a product in which a product ID is registered in the product database DB3. The vector of similar products is a vector generated from the content contained in the product detail page on which similar products are posted, using the same method as the method of vectorizing recommended products, and has the same number of dimensions as the vector of recommended products. It is assumed that it is a vector having (the number of elements).

Then, the optimization processing unit 104 selects a vector of similar products whose similarity to the vector of the recommended product is equal to or higher than the threshold value from the vectors of a plurality of similar products, and selects the similar product that is the source of the selected vector. It is a recommended product to be posted in the product posting frame.

FIG. 12 is a diagram for explaining a method of determining a product similar to a recommended product. For example, the first recommended items database DB4, to the target product ID, if the product ID of the plurality of recommended items is associated, the optimization processor 104, compared Shipping I _TA with Shipping ID Te derives a其's product vectors of a plurality of recommended items I ₁ ~I _k the product ID is associated, the similarity between vectors of similar products, vectors of similar products in which the similarity is equal to or more than a threshold Select. In the illustrated example, there are three similar product vectors whose similarity is equal to or higher than the threshold value for the product vector _{of the recommended product I 1, and the similar products corresponding to each of the three similar product vectors.} I _{1 ',} I _2', the _{I 3} ', to determine the recommended items to be posted on the goods posted frame. At this time, the optimization processing unit 104 may set _{each weight of the similar products I 1} ′, I ₂ ′, and I ₃ ′ as the degree of similarity. As a result, _{among the products I 1 to} I _k and similar products I ₁ ', I ₂ ', and I ₃ ', the products with the heaviest weight are preferentially placed in the product posting frame.

In this way, in the first recommended product database DB4, a product similar to the recommended product in which the product ID is associated with the target product ID is set as a recommended product to be posted in the product posting frame, so that the product can be purchased together in the past. In addition to making other products that have been sold or put in the shopping cart at the same time as recommended products, products similar to the recommended products can also be made recommended products, so that consumers can shop. It is possible to widely recommend products that are likely to take certain actions such as putting them in a cart or purchasing them.

Further, the optimization processing unit 104 may determine the weight of the recommended product by machine learning using a support vector machine or a neural network. For example, the optimization processing unit 104 uses the number of product reviews, the degree of matching of product titles, the product price, and the like as the characteristics, and the number of reviews is large, the evaluation of the reviews is high, and the degree of matching of product titles is large. , The parameters of a predetermined learning model are learned so that the lower the product price, the smaller the weight of the product. The optimization processing unit 104 determines the weight of the recommended product by inputting the feature of the recommended product in which the product ID is associated with the target product ID in the first recommended product database DB4 into the learning model learned in advance. .. As a result, a product that easily causes a predetermined action can be selected as a recommended product.

In the above-described embodiment, when the information processing apparatus 100 receives a predetermined request, the target product indicated by the target product ID included in the predetermined request is purchased together with other products in the past. However, this is not limited to the case where other products purchased at the same time or other products added to the shopping cart at the same time are recommended as recommended products when they are added to the shopping cart at the same time.

For example, the information processing device 100 may provide the terminal device 10 with advertising content related to other products purchased at the same time or other products put into the shopping cart at the same time as a response to a predetermined request.

FIG. 13 is a diagram showing an example of a web page to which advertising content is delivered. For example, the optimization processing unit 104 is a recommended product in which the product ID is associated with the target product ID in the first recommended product database DB 4, that is, a product included in the category of the target product ID or a category combined with the category. It is decided to place the advertisement content that introduces the product in the predetermined space (the area of R6 in the figure). The advertising content may be, for example, a still image or a moving image. Then, the optimization processing unit 104 extracts the advertisement content to be posted in the predetermined placement frame from the predetermined database (not shown) in which the advertisement content is stored, and outputs this to the distribution management unit 102. In response to this, the distribution management unit 102 transmits the advertisement content to the terminal device 10 as a response to a predetermined request.

According to the embodiment described above, the behavior history of a plurality of users who have reached a predetermined behavior for a product sold via a network is one for each session divided based on a series of user behaviors. Generates a plurality of first aggregated log data summarized in, and generates a plurality of second aggregated log data in which products included in each of the plurality of first aggregated log data are grouped into one for each category. In order to generate recommended product candidate list data that summarizes a plurality of products that the user has reached a predetermined action in the same session among the products included in each of the second aggregated log data of the above, for example. When a user who is a consumer browses a product as a predetermined action, it is easy for the user to purchase a product associated with the product browsed by the user in the recommended product candidate list data (predetermined action). Can be recommended as a product (easy to wake up).

For example, in the above-described embodiment, when a plurality of products are purchased at once (during the same session), the first aggregated log data is collected for each category as the second aggregated log data, but they are purchased at the same time. For the products that are in the same category, the second aggregated log data is recombined even if the products are in different categories. As a result, for example, when a user purchases a product in the category of "shirt" and a product in the category of "shoes" at the same time, a product in the category of "shirt" and a product in the category of "shoes" should be used. Where they are separated from each other, they are combined into one (the sessions are merged). As a result, products in both categories are treated as products that are related to each other beyond the boundaries of the categories. Therefore, for example, a user who is a consumer may put products in the "fashion" category into a shopping cart. , If you make a purchase, you can recommend products in the category merged into that category (for example, "food") in addition to products in the "fashion" category, and products that are easy for users to take certain actions. Can be recommended more accurately.

<Hardware configuration>
The information processing apparatus 100 of the above-described embodiment is realized by, for example, a hardware configuration as shown in FIG. FIG. 14 is a diagram showing an example of the hardware configuration of the information processing apparatus 100 of the embodiment.

The information processing device 100 includes NIC100-1, CPU100-2, RAM100-3, ROM100-4, a secondary storage device 100-5 such as a flash memory or an HDD, and a drive device 100-6 having an internal bus or a dedicated communication line. It is configured to be interconnected by. A portable storage medium such as an optical disk is mounted on the drive device 100-6. A program stored in a portable storage medium mounted on the secondary storage device 100-5 or the drive device 100-6 is expanded in the RAM 100-3 by a DMA controller (not shown) or the like, and executed by the CPU 100-2. As a result, each component of the distribution management unit 102, the optimization processing unit 104, the database update unit 106, and the database generation processing unit 110 is realized. The programs referenced by these components may be downloaded from other devices via the network NW.

Although the embodiments for carrying out the present invention have been described above using the embodiments, the present invention is not limited to these embodiments, and various modifications and substitutions are made without departing from the gist of the present invention. Can be added.

1 ... Information processing system, 10 ... Terminal device, 20 ... Service providing device, 100 ... Information processing device, 102 ... Distribution management unit, 104 ... Optimization processing unit, 106 ... Database update unit, 110 ... Database generation processing unit, 112 … First data generation unit, 114… specific history extraction unit, 116… second data generation unit

Claims (8)

One or more first data that summarizes the behavior history of a user who has reached a predetermined behavior for a product or service sold via a network for each processing reference period based on the series of behaviors of the user. The first generator to generate and
A second generation unit that generates one or more second data in which the goods or services included in each of the one or more first data generated by the first generation unit are grouped into one for each category. ,
Among the goods or services included in each of the one or more second data generated by the second generation unit, a plurality of the goods or services in which the user has reached a predetermined action within the same processing reference period. A third generation unit that generates one or more third data that combine each other into one,
The acquisition unit that acquires the identification information of the product or service that led to the user's predetermined behavior,
A recommendation processing unit that recommends the product or service acquired by the acquisition unit together with the product or service indicated by the identification information to the user who has taken the predetermined action.
Information processing device equipped with. The third generation unit is
Combining each category of a plurality of the goods or services that the user has achieved a predetermined action within the same processing reference period into one.
Collecting the goods or services classified into the combined categories to generate the third data.
The information processing apparatus according to claim 1. The prescribed actions include purchasing goods or services.
The third generation unit is a plurality of the products or services purchased by the user within the same processing reference period from the products or services included in each of the second data generated by the second generation unit. Generate the third data that combines services into one,
The information processing apparatus according to claim 1 or 2. The recommendation processing unit is a product or service that has a large number of predetermined actions within the same processing reference period among a plurality of products or services that are combined into one as the third data together with the product or service indicated by the identification information. Or, the service is given priority and recommended to the user who has taken the predetermined action.
The information processing apparatus according to any one of claims 1 to 3. Further, the recommendation processing unit further performs the product or service similar to the product or service collectively as the third data together with the product or service indicated by the identification information acquired by the acquisition unit. Recommended to the user who caused the problem,
The information processing apparatus according to any one of claims 1 to 4. The acquisition unit that acquires the identification information of the product or service that led to the user's predetermined behavior,
A recommendation processing unit that recommends an advertisement related to a product or service acquired by the acquisition unit together with the product or service indicated by the identification information as the third data to a user who has taken the predetermined action. Further prepare,
The information processing apparatus according to any one of claims 1 to 5. The computer
One or more first data that summarizes the behavior history of a user who has reached a predetermined behavior for a product or service sold via a network for each processing reference period based on the series of behaviors of the user. Generate and
One or more second data in which the goods or services included in each of the one or more generated first data are grouped into one for each category are generated.
Among the products or services included in each of the generated one or more second data, a plurality of the products or services in which the user has reached a predetermined action within the same processing reference period are grouped together. Generate one or more third data ,
Acquires the identification information of the product or service that led to the user's predetermined behavior,
The goods or services combined as the third data together with the goods or services indicated by the acquired identification information are recommended to the user who has taken the predetermined action.
Information processing method. On the computer
One or more first data that summarizes the behavior history of a user who has reached a predetermined behavior for a product or service sold via a network for each processing reference period based on the series of behaviors of the user. The process to generate and
A process of generating one or more second data in which the goods or services included in each of the one or more generated first data are grouped into one for each category.
Among the products or services included in each of the generated one or more second data, a plurality of the products or services in which the user has reached a predetermined action within the same processing reference period are grouped together. The process of generating one or more third data,
The process of acquiring the identification information of the product or service that led to the user's predetermined behavior,
A process of recommending a product or service integrated as the third data together with the acquired product or service indicated by the identification information to a user who has taken the predetermined action.
A program to execute.

Images