JP2020154880A - Calculation device, calculation method, and calculation program - Google Patents

Abstract

To provide a calculation device capable of providing information more strongly attracts a user's interest, a calculation method, and a calculation program.SOLUTION: A calculation device comprises an acquisition unit which acquires action information indicating actions of a user; and a calculation unit which calculates a score indicating the possibility of purchasing a transaction object by the user on the basis of whether action information indicating browsing of the transaction object is included. In addition, the calculation device includes a determination unit which determines whether to display information on the transaction object on content of a domain different from an electronic mall on the basis of the score calculated by the calculation unit.SELECTED DRAWING: Figure 10

Description

The present invention relates to a calculation device, a calculation method, and a calculation program.

Services using the network are actively performed. For example, a technology for recommending a transaction target according to a user's attribute is known on an EC (Electronic Commerce) site or the like that realizes electronic commerce of the transaction target such as various products and services.

Japanese Unexamined Patent Publication No. 2013-205875

However, it may not be said that the above-mentioned conventional technology provides information that is of interest to the user.

For example, in the above-mentioned conventional technology, since the transaction target is only recommended according to the attribute of the user registered in advance, the information of the transaction target desired by the user is provided when making a recommendation to the user. It may not be possible to say that.

The present application has been made in view of the above, and an object of the present application is a calculation device, a calculation method, and a calculation program capable of providing information that is more strongly interesting to the user.

The calculation device according to the present application is based on an acquisition unit that acquires behavior information indicating the user's behavior, and whether or not the behavior information acquired by the acquisition unit includes behavior information indicating browsing of a transaction target. It is characterized by having a calculation unit for calculating a score indicating the possibility that the user purchases the transaction target.

According to one aspect of the embodiment, there is an effect that the user can provide information that is more strongly interesting.

FIG. 1 is a diagram showing an example of a determination process executed by the calculation device according to the first embodiment. FIG. 2 is a diagram showing an example of a selection process for selecting selection logic executed by the calculation device according to the first embodiment. FIG. 3 is a diagram showing an example of an extraction process for extracting a log executed by the calculation device according to the first embodiment. FIG. 4 is a diagram showing a configuration example of the calculation device according to the first embodiment. FIG. 5 is a diagram showing an example of a log information storage unit according to the first embodiment. FIG. 6 is a diagram showing an example of the selection logic storage unit according to the first embodiment. FIG. 7 is a flowchart showing an example of the flow of the determination process executed by the calculation device according to the first embodiment. FIG. 8 is a flowchart showing an example of a flow of log estimation processing executed by the calculation device according to the first embodiment. FIG. 9 is a diagram showing an example of a determination process executed by the calculation device according to the second embodiment. FIG. 10 is a diagram showing a configuration example of the calculation device according to the second embodiment. FIG. 11 is a flowchart showing an example of a flow of content provision processing executed by the calculation device according to the second embodiment. FIG. 12 is a hardware configuration diagram showing an example of a computer that realizes the function of the calculation device.

Hereinafter, a calculation device, a calculation method, and a mode for carrying out the calculation program according to the present application (hereinafter, referred to as “the embodiment”) will be described in detail with reference to the drawings. The calculation device, calculation method, and calculation program according to the present application are not limited by this embodiment. In addition, each embodiment can be appropriately combined as long as the processing contents do not contradict each other. Further, in each of the following embodiments, the same parts are designated by the same reference numerals, and duplicate description is omitted.

[1. First Embodiment]
[1-1. An example of the determination process shown by the calculation device]
An example of the determination process executed by the calculation device 100 according to the first embodiment will be described with reference to FIG. FIG. 1 is a diagram showing an example of a determination process executed by the calculation device 100 according to the first embodiment. Specifically, the calculation device 100 uses a method selected from a plurality of methods for determining a transaction target for providing information to the user from the behavior information indicating the user's behavior, and uses the behavior information. , Determine the transaction target that provides information to the user.

The behavior information referred to here is information about the behavior performed by the user. For example, the behavior information is information on the behavior of the user performed on the shopping site and the EC (Electric Commerce) site. Further, for example, the behavior information is information regarding the behavior of the user performed on the auction site.

Further, for example, the behavior information includes a search log searched by the user, a browsing history of the content browsed by the user, a purchase history related to the product purchased by the user, and the like. Such behavioral information can be realized by associating it with information (for example, cookie information) that identifies a terminal device used by a user.

In addition, the transaction target referred to here is a product or service. For example, a service is not only a service used by a user through an actual store, but also a service that is ordered by the user via a network and is directly provided to the user, or an order is placed. It includes services provided to users other than users, services provided via networks, and the like. Further, in the following, an example will be described assuming that the transaction target is a product.

As shown in FIG. 1, the calculation system 1 includes a terminal device 10 and a calculation device 100. The terminal device 10 and the calculation device 100 are communicably connected by wire or wirelessly via a predetermined communication network (not shown). The calculation system 1 shown in FIG. 1 may include a plurality of terminal devices 10 and a plurality of calculation devices 100.

The terminal device 10 according to the first embodiment is an information processing device used by a user who accesses contents such as a web page displayed on a browser and contents for an application. For example, the terminal device 10 is a desktop PC (Personal Computer), a notebook PC, a tablet terminal, a mobile phone, a PDA (Personal Digital Assistant), or the like.

In the example of FIG. 1, a case where the terminal device 10 is a smartphone used by a user is shown. In the following, the terminal device 10 may be referred to as a user. That is, in the following, the user can be read as the terminal device 10. Further, in the following, when it is not necessary to distinguish the terminal device 10 used by each user, it may be referred to as the terminal device 10.

The calculation device 100 according to the first embodiment is realized by, for example, a server device or the like. Specifically, the calculation device 100 acquires behavior information indicating the behavior of the user. Then, the calculation device 100 selects one of a plurality of methods for determining a transaction target for providing information to the user from the behavior information based on the behavior indicated by the behavior information. Then, the calculation device 100 uses this method to determine the transaction target that provides the information to the user from the behavior information.

Hereinafter, an example of the determination process by the calculation device 100 will be described along the flow with reference to FIG.

First, as shown in FIG. 1, the calculation device 100 acquires a log, which is user behavior information (step S1). For example, in the example of FIG. 1, the calculation device 100 uses the user as a log to obtain a search log searched by the user, a browsing history of the content browsed by the user, and a purchase history of the product purchased by the user. Obtained from the terminal device 10 used.

Subsequently, the calculation device 100 generates a learning model M1 that has learned the characteristics of the log and the characteristics of the product related to the log (step S2). For example, the calculation device 100 uses conventional techniques of machine learning methods such as logistic regression, SVM (Support Vector Machine), and deep learning to obtain the characteristics of a predetermined user's log and products related to the predetermined user's log. A learning model M1 is generated based on the features.

For example, the calculation device 100 uses the conventional technique of the machine learning method as a log of a predetermined user, which includes features such as a content type, information about a domain indicated by the content, a date and time indicated by a search log and a purchase history, and a predetermined use. As a feature of the product related to the log of the person, the learning model M1 is generated based on the information about the product, the information about the category of the product indicated by the product, and the like.

Then, the calculation device 100 refers to the user specified by the target user ID (Identifier) "U100" (hereinafter, the user specified by the user ID "UN" is "user UN (N is arbitrary). (Numerical value) ”is sometimes written) from the terminal device 10 (step S3). That is, the calculation device 100 acquires the log of the target user, which is information necessary for selecting a product that stimulates the purchase motivation of the target user.

Subsequently, the calculation device 100 selects one of the selection logics for selecting the product based on whether or not the log type satisfies a predetermined condition (step S4). That is, the calculation device 100 is a selection logic for selecting a product that stimulates the purchase motivation of the user, and is a condition or the like that the user's log satisfies from the selection logic that selects the product by different methods. Based on this, select the selection logic that has the highest probability of selecting the product that is most selected by the user. Specifically, the calculation device 100 provides information to the user from a transaction target related to the action indicated by the action information based on the relationship of a plurality of transaction targets viewed by the same user. Select one of the multiple methods, including at least the relevance method, which is the method for determining.

Further, specifically, the calculation device 100 includes at least a plurality of search methods that are methods for determining a transaction target that provides information to the user from a transaction target that is related to the search query input by the user. Select one of the methods from.

More specifically, the calculation device 100 selects the relevance method when the number or ratio of the behavior information indicating the browsing of the transaction target among the behavior information acquired by the transaction target satisfies a predetermined condition. More specifically, the calculation device 100 selects a search method when the number or ratio of the behavior information indicating the browsing of the transaction target does not satisfy the predetermined condition among the behavior information acquired by the transaction target. To do.

Here, the selection process for selecting the selection logic will be described with reference to FIG. FIG. 2 is a diagram showing an example of a selection process for selecting selection logic executed by the calculation device according to the first embodiment. Further, an example of the selection process shown in FIG. 2 is a process corresponding to step S4 in FIG. Table TB1 is a table showing the logs of the target users. For example, as shown in FIG. 2, the user has log information. Specifically, the log information corresponding to the user U1 is a log ID, a type, a date and time, a product, and a category.

For example, the type corresponding to the log specified by the log ID "L1" (hereinafter, the log specified by the log ID "LN" may be referred to as "log LN (N is an arbitrary numerical value)") , Indicates shopping browsing. Further, for example, the date and time corresponding to the log L1 indicates the date and time T1. Further, for example, the product corresponding to the log L1 indicates the product MD1. Further, for example, the product category corresponding to the log L1 indicates the category CA1.

Further, for example, the type corresponding to the log L2 indicates a shopping search. Further, for example, the date and time corresponding to the log L2 indicates the date and time T2. Further, for example, the product corresponding to the log L2 indicates the product MD2. Further, for example, the product category corresponding to the log L2 indicates the category CA2.

Further, for example, the type corresponding to the log L3 indicates an auction search. Further, for example, the date and time corresponding to the log L3 indicates the date and time T3. Further, for example, the product corresponding to the log L3 indicates the product MD3. Further, for example, the category of the product corresponding to the log L3 indicates the category CA3.

Further, for example, the type corresponding to the log L4 indicates commerce. Here, commerce is intended for commerce. For example, a product intended for commerce is a search query that includes a product name, etc., a product that corresponds to the search query, and a link to product information or sales-related content is set. is there. Further, for example, the date and time corresponding to the log L4 indicates the date and time T4. Further, for example, the product corresponding to the log L4 indicates the product MD4. Further, for example, the product category corresponding to the log L4 indicates the category CA4.

Further, for example, the type corresponding to the log L5 indicates a link. Here, the link is a search log containing links related to other different shopping services. Further, for example, the date and time corresponding to the log L5 indicates the date and time T5. Further, for example, the product corresponding to the log L5 indicates the product MD5. Further, for example, the category of the product corresponding to the log L5 indicates the category CA5.

Table TB2 is a table relating to selection logic. For example, as shown in FIG. 2, in Table TB2, the selection logic specified by the logic ID “LG1” (hereinafter, the log specified by the logic ID “LGN” is referred to as “logic LGN (N is an arbitrary numerical value)”. It has information about the selection logic for each (may be written as). Specifically, the information regarding the selection logic is a type and a condition.

For example, the type corresponding to logic LG1 indicates collaborative filtering. Here, the collaborative filtering is, for example, a method of calculating the similarity indicating the similarity between users and selecting a product based on the relevance. For example, the calculation device 100 stores in the storage unit a product that is likely to be viewed by the same user in association with the log of each user. Then, the calculation device 100 identifies the product browsed by the target user from the browsing history of the target user, and determines the product as a product that proposes the product associated with the specified product. When a plurality of products to be proposed are determined, the calculation device 100 determines one by a conventional technique such as various selection logics. Further, for example, the condition corresponding to the logic LG1 indicates the condition CON1.

Further, for example, the type corresponding to logic LG2 indicates history selection. Here, the history selection is, for example, a method of selecting a product with a large number of searches, and for example, a method of identifying a product whose number of searches is equal to or greater than a predetermined threshold value. In this case, the calculation device 100 identifies a product whose number of searches is equal to or greater than a predetermined threshold value from the log of each user, and determines the product as a product that proposes the specified product. Further, for example, the condition corresponding to the logic LG2 indicates the condition CON2.

Further, for example, the type corresponding to logic LG3 indicates a related selection. Here, the related selection is, for example, a method of selecting a product included in the same category as a product purchased in the past by the user, for example, in the same category as the product purchased in the past by the user. This is a method of identifying products that are frequently purchased by other users among the included products. In this case, the calculation device 100 identifies from the logs of each user the products that are included in the same category as the products that were purchased by the user in the past and that are frequently purchased by other users. , Determine as a product to propose the specified product. Further, for example, the condition corresponding to the logic LG3 indicates the condition CON3.

In the example of FIG. 2, the calculation device 100 calculates a score indicating the possibility of the user selecting the transaction target determined by the method for each method based on the behavior, and selects the method having the highest calculated score. To do. As shown in FIG. 2, the calculation device 100 calculates a score according to the log type for each product selection logic, and selects the selection logic having the highest score (step S21). As a result of such processing, the calculation device 100 selects an appropriate selection logic, so that the user can select a product that is more strongly interested.

For example, in the example of FIG. 2, the calculation device 100 calculates a score for each selection logic included in the table TB2. For example, when the number of shopping views included in the user's log is equal to or greater than a predetermined threshold value, the score of the selection logic using the number of shopping views is increased. In this case, the collaborative filtering score is increased. In other words, the collaborative filtering condition includes a condition in which the score value increases as the number of shopping views increases. For example, in the example of FIG. 2, the calculation device 100 calculates that the score corresponding to the collaborative filtering associated with the logic LG1 is 0.5 based on the log included in the table TB1. Further, for example, the calculation device 100 calculates that the score corresponding to the history selection associated with the logic LG2 is 0.2 based on the log included in the table TB1. Further, for example, the calculation device 100 calculates that the score corresponding to the related selection associated with the logic LG3 is 0.1 based on the log included in the table TB1. Then, the calculation device 100 selects the logic LG1 having the highest score from the plurality of selection logics.

For example, if information about a product viewed in the past is provided to the user, there is a high possibility that the product will be selected by the user. For example, when the number of views of the product included in the user's log in the past is equal to or greater than a predetermined threshold value, the score of the selection logic using the number of views is increased. In this case, the score of history selection is increased. In other words, the history selection includes a condition that the score value increases as the number of views of the product viewed in the past increases. Further, for example, among the products included in the user's log that are included in the same category as the products viewed in the past, the number of purchases is used when the number of purchases by other users is equal to or more than a predetermined threshold value. Increases the selection logic score. In this case, the score of the related selection is increased. In other words, the condition of the related selection includes a condition that the score value becomes higher as the number of times of purchase by other users is larger among the products included in the same category as the products viewed in the past. Further, as the condition corresponding to each selection logic, a condition for reducing the weight of the log over time may be set.

The calculation device 100 may determine the product by using the selection logic having the highest score (step S22). For example, it is assumed that collaborative filtering is selected as the selection logic. In this case, the calculation device 100 may determine the product to be provided to the target user by using collaborative filtering. For example, in the example of FIG. 2, the calculation device 100 determines the product MD1 from the log L10 by using collaborative filtering. Further, for example, in the example of FIG. 2, the calculation device 100 determines the product MD5 from the log L20 by using collaborative filtering.

Further, in the example shown in FIG. 2, the date and time are represented by an abstract code such as "T1", but the date and time may be a specific numerical value or the like. Further, although the product is represented by an abstract code such as "MD1", the product may be information regarding a specific product or service name or the like. Further, although the category is represented by an abstract code such as "CA1", the category may be information related to a specific name or the like. Further, although the condition is expressed by an abstract code such as "CON1", the condition may be a concrete numerical value or the like.

Returning to FIG. 1, an example of the determination process according to the first embodiment will be described. The calculation device 100 estimates a useful log using the learning model M1 (step S5). Subsequently, the calculation device 100 uses the useful log to determine the product (step S6).

Here, the estimation process for estimating the log will be described with reference to FIG. FIG. 3 is a diagram showing an example of an extraction process for extracting a log executed by the calculation device according to the first embodiment. Further, an example of the estimation process shown in FIG. 3 is a process corresponding to steps S5 and S6 of FIG. Table TB3 is a table showing logs corresponding to each of a plurality of users. Specifically, the log information corresponding to each of the plurality of users is the log ID, the user ID, the type, the date and time, the product, the category, and the number of clicks on the content.

For example, the user ID corresponding to the log L1 indicates U1. Further, for example, the type corresponding to the log L1 indicates shopping browsing. Further, for example, the date and time corresponding to the log L1 indicates the date and time T1. Further, for example, the product corresponding to the log L1 indicates the product MD1. Further, for example, the product category corresponding to the log L1 indicates the category CA1. Further, for example, the number of clicks corresponding to the log L1 indicates 1.

Further, for example, the user ID corresponding to the log L2 indicates U1. Further, for example, the type corresponding to the log L2 indicates a shopping search. Further, for example, the date and time corresponding to the log L2 indicates the date and time T2. Further, for example, the product corresponding to the log L2 indicates the product MD2. Further, for example, the product category corresponding to the log L2 indicates the category CA2. Further, for example, the number of clicks corresponding to the log L2 indicates 0.

Further, for example, the user ID corresponding to the log L3 indicates U2. Further, for example, the type corresponding to the log L3 indicates an auction search. Further, for example, the date and time corresponding to the log L3 indicates the date and time T3. Further, for example, the product corresponding to the log L3 indicates the product MD3. Further, for example, the category of the product corresponding to the log L3 indicates the category CA3. Further, for example, the number of clicks corresponding to the log L3 indicates 1.

Further, for example, the user ID corresponding to the log L4 indicates U3. Further, for example, the type corresponding to the log L4 indicates commerce. Further, for example, the date and time corresponding to the log L4 indicates the date and time T4. Further, for example, the product corresponding to the log L4 indicates the product MD4. Further, for example, the product category corresponding to the log L4 indicates the category CA4. Further, for example, the number of clicks corresponding to the log L4 indicates 0.

Further, for example, the user ID corresponding to the log L5 indicates U4. Further, for example, the type corresponding to the log L5 indicates a link. Here, the link is a link related to another different shopping service. Further, for example, the date and time corresponding to the log L5 indicates the date and time T5. Further, for example, the product corresponding to the log L5 indicates the product MD5. Further, for example, the category of the product corresponding to the log L5 indicates the category CA5. Further, for example, the number of clicks corresponding to the log L5 indicates 1.

In the example of FIG. 3, the calculation device 100 uses a model that learns the relationship between the degree of interest of the user in the transaction target related to the behavior indicated by the behavior information and the characteristics of the behavior information for each behavior information. In addition, the degree of interest of the user is estimated. More specifically, the calculation device 100 estimates the degree of interest of the user by using a model that learns the relationship between the type of action indicated by the behavior information and the degree of interest of the user as a feature. More specifically, the calculation device 100 uses a model that learns the relationship between whether or not the behavior indicated by the behavior information is browsing of a transaction target related to the behavior and the degree of interest of the user. Estimate the degree of interest in.

More specifically, the calculation device 100 uses a model that learns the relationship between whether or not the behavior indicated by the behavior information is a web search related to the transaction target and the degree of interest of the user. Estimate the degree of interest. More specifically, the calculation device 100 determines whether or not the action indicated by the action information is a web search of information related to the transaction target, and whether or not the action indicated by the action information is a web search in which the information of the transaction target is displayed in the search result. The degree of interest of the user is estimated using the model that learned the relationship with the degree of interest of.

In addition, the calculation device 100 has learned the relationship between whether or not the behavior indicated by the behavior information is a web search displaying a search result including a link to information related to the transaction target and the degree of interest of the user. The model is used to estimate the degree of interest of the user.

As shown in FIG. 3, the calculation device 100 generates a learning model M1 that has learned the relationship between the log type and the CTR (Click Through Rate) for the content including the product corresponding to the log. (Step S31). For example, the calculation device 100 generates the learning model M1 by learning the relationship between the input information such as the log type, the date and time, the product, and the category, and the CTR calculated from the number of clicks on the content. To explain a more specific example, when the calculation device 100 inputs the log type, the date and time, the product, and the category, the calculation device 100 shows the input log type, the date and time, the product, and the category. The learning model M1 is trained so as to output the CTR value calculated from.

Then, the calculation device 100 estimates the CTR of each log using the learning model M1 (step S32). For example, the calculation device 100 outputs a score when a log to be predicted is input based on the learning model M1. Then, the calculation device 100 estimates the CTR corresponding to each log by converting the score output from the learning model M1 into a probability value by using Sigmoid fitting or the like.

For example, in the example of FIG. 3, the calculation device 100 estimates the CTR to be “0.2” from the information about the log L21 by using the learning model M1. Further, for example, the calculation device 100 estimates the CTR to be "0.002" from the information regarding the log L22 by using the learning model M1. Further, for example, the calculation device 100 estimates the CTR as "0.0001" from the information regarding the log L23 by using the learning model M1.

Further, for example, the calculation device 100 estimates the CTR to be "0.01" from the information regarding the log L24 by using the learning model M1. Further, for example, the calculation device 100 estimates the CTR to be "0.05" from the information regarding the log L25 by using the learning model M1.

Then, the calculation device 100 determines the product using the log in which the CTR is equal to or higher than a predetermined threshold value among the CTRs estimated using the learning model M1 (step S33). That is, the calculation device 100 selects a product using a log estimated to have a high CTR, that is, a log estimated to more strongly reflect the intention of the user. For example, it is assumed that the logs whose CTR is equal to or higher than a predetermined threshold are the log L10 and the log L20. It is also assumed that collaborative filtering is selected as the selection logic. In this case, the calculation device 100 uses collaborative filtering to determine the product to be provided to the target user based on the log in which the CTR is equal to or higher than a predetermined threshold value. As a result of such processing, the calculation device 100 selects the product from the log that more strongly reflects the intention of the user, so that the product that the user is more strongly interested in can be selected as the proposal target.

For example, in the example of FIG. 3, the calculation device 100 determines the product MD1 from the information about the log L10 by using selection logic such as collaborative filtering. Further, for example, in the example of FIG. 3, the calculation device 100 determines the product MD5 from the information regarding the log L20 by using selection logic such as collaborative filtering.

In the example shown in FIG. 3, the date and time are represented by an abstract code such as "T1", but the date and time may be a specific numerical value or the like. Further, although the product is represented by an abstract code such as "MD1", the product may be information regarding a specific product or service name or the like. Further, although the category is represented by an abstract code such as "CA1", the category may be information related to a specific name or the like.

Returning to FIG. 1, an example of the determination process according to the first embodiment will be described. The calculation device 100 provides the user U100 with content including information about the product (step S7). For example, the calculation device 100 provides the user U100 with content CR1 including information about a plurality of products. For example, in the example of FIG. 1, the calculation device 100 includes the product A, the price of the product A "1200 yen", the price of the product B and the product B "1500 yen", the price of the product C, and the price of the product C " The content CR1 in which "1700 yen" is displayed at a predetermined location is provided to the user U100.

In the example of FIG. 1, the calculation device 100 provides the user U100 with the content CR1 including the information about the product A, the information about the product B, and the information about the product C. It does not have to be limited. Specifically, the calculation device 100 may provide the user with content including information on three or more types of products. For example, the calculation device 100 provides the user with content including information about the product A, information about the product B, information about the product C, information about the product D, information about the product E, and information about the product F. May be done.

In this way, the calculation device 100 according to the first embodiment acquires the behavior information indicating the behavior of the user. Then, the calculation device 100 selects one of a plurality of methods for determining a transaction target for providing information to the user from the behavior information based on the behavior indicated by the behavior information. Then, the calculation device 100 uses this method to determine the transaction target that provides the information to the user from the behavior information. As a result, the calculation device 100 according to the first embodiment uses the method to determine the transaction target for providing the information to the user from the behavior information, so that the information that the user is more strongly interested in is obtained. Can be provided.

This point will be described. Explaining with reference to the example of FIG. 1, the calculation device 100 estimates a useful log by using the learning model M1 that has learned the characteristics of the log and the characteristics of the product related to the log. Then, the calculation device 100 determines a product from such a useful log by using selection logic such as collaborative filtering. As a result, the calculation device 100 can determine a product suitable for the user, so that the product that the user is more strongly interested in can be determined with high accuracy. From this, the calculation device 100 can provide information that is more strongly interesting to the user, and thus can provide a better service.

[1-2. Configuration of calculation device]
Next, the configuration of the calculation device 100 according to the first embodiment will be described with reference to FIG. FIG. 4 is a diagram showing a configuration example of the calculation device 100 according to the first embodiment. As shown in FIG. 4, the calculation device 100 includes a communication unit 110, a storage unit 120, and a control unit 130.

(About communication unit 110)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card) or the like. Then, the communication unit 110 is connected to the network by wire or wirelessly, and transmits / receives information to / from the terminal device 10.

(About storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory (Flash Memory), or a storage device such as a hard disk or an optical disk. The storage unit 120 includes a log information storage unit 121, a learning model 122, and a selection logic storage unit 123.

(About log information storage unit 121)
The log information storage unit 121 according to the first embodiment stores information related to the user's log. Here, FIG. 5 shows an example of the log information storage unit 121 according to the first embodiment. In the example shown in FIG. 5, the log information storage unit 121 has items such as "log ID", "type", "date and time", "product", and "category".

The "log ID" is an identifier that identifies the log. The "type" is information regarding the type of log associated with the "log ID". The "date and time" is a date and time associated with the "log ID", and is information regarding the date and time when the log is stored.

The "product" is information about the product associated with the "log ID". The "category" is a category associated with the "log ID" and is information about the product category. For example, suppose the product is beer. In this case, the category is liquor.

For example, in FIG. 5, the type of "L1" identified by the log ID is "shopping browsing", the date and time is "T1", the product is "MD1", and the category is "CA1". In the example shown in FIG. 5, the date and time are represented by an abstract code such as "T1", but the date and time may be a specific numerical value or the like. Further, although the product is represented by an abstract code such as "MD1", the product may be information regarding a specific product name or the like. Further, although the category is represented by an abstract code such as "CA1", the category may be information related to a specific name or the like.

(About the selection logic storage unit 123)
The selection logic storage unit 123 according to the first embodiment stores information regarding the selection logic. Here, FIG. 6 shows an example of the selection logic storage unit 123 according to the first embodiment. In the example shown in FIG. 6, the selection logic storage unit 123 has items such as "logic ID", "type", "condition", and "score".

The "logic ID" is an identifier that identifies the logic. The "type" is information regarding the type of selection logic associated with the "logic ID". The "condition" is information about a condition related to the selection logic associated with the "logic ID". The "score" is information about the score according to the log type for each product selection logic associated with the "logic ID".

For example, in FIG. 6, “LG1” identified by the logic ID has a type of “collaborative filtering”, a condition of “CON1”, and a score of “0.5”. In the example shown in FIG. 6, the condition is represented by an abstract code such as "CON1", but the condition may be a specific numerical value or the like.

(About control unit 130)
The control unit 130 is a controller, and for example, various programs (as an example of a calculation program) stored in a storage device inside the calculation device 100 by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. (Equivalent) is realized by executing RAM as a work area. Further, the control unit 130 is a controller, and is realized by, for example, an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

As shown in FIG. 4, the control unit 130 includes an acquisition unit 131, a generation unit 132, a selection unit 133, an estimation unit 134, a determination unit 135, and a provision unit 136, and the information described below. Realize or execute the function or action of processing. The internal configuration of the control unit 130 is not limited to the configuration shown in FIG. 4, and may be another configuration as long as it is a configuration for performing information processing described later. Further, the connection relationship of each processing unit included in the control unit 130 is not limited to the connection relationship shown in FIG. 4, and may be another connection relationship.

(About acquisition unit 131)
The acquisition unit 131 acquires various types of information. Specifically, the acquisition unit 131 acquires behavior information indicating the behavior of the user. For example, in the example of FIG. 1, the acquisition unit 131 records the search log searched by the user, the browsing history of the content browsed by the user, and the purchasing history of the product purchased by the user by the user. Obtained from the terminal device 10 used. Then, the acquisition unit 131 stores such a log in the log information storage unit 121.

(About generator 132)
The generation unit 132 generates a learning model based on the log stored in the log information storage unit 121. Specifically, the generation unit 132 generates a learning model 122 that learns the characteristics of the log and the characteristics of the product related to the log. Further, specifically, the generation unit 132 generates a learning model 122 that has learned the relationship between the log type and the content including the product corresponding to the log with the CTR.

For example, the generation unit 132 is based on the characteristics of a predetermined user's log and the characteristics of a product related to the predetermined user's log by using conventional techniques of machine learning methods such as logistic regression, SVM, and deep learning. , Generate a learning model 122.

For example, according to the conventional technique of the machine learning method, the generation unit 132 has features such as the type of content, information about the domain indicated by the content, the date and time indicated by the search log and the purchase history, and the predetermined use as the log of the predetermined user. As a feature of the product related to the log of the person, the learning model 122 is generated based on the information about the product, the information about the category of the product indicated by the product, and the like.

For example, the generation unit 132 generates the learning model 122 by learning the relationship between the input information such as the log type, the date and time, the product, and the category, and the CTR calculated from the number of clicks on the content.

(About selection unit 133)
The selection unit 133 selects one of a plurality of methods for determining a transaction target for providing information to the user from the behavior information based on the behavior indicated by the behavior information acquired by the acquisition unit 131. To do. Specifically, the selection unit 133 calculates a score indicating the possibility of the user selecting the transaction target determined by the method for each method based on the behavior, and selects the method having the highest calculated score. .. Further, specifically, the selection unit 133 selects one of a plurality of methods based on whether or not the action indicated by the action information satisfies a predetermined condition.

More specifically, the selection unit 133 provides information to the user from the transaction target related to the action indicated by the action information based on the relationship of a plurality of transaction targets viewed by the same user. Select one of the multiple methods, including at least the relevance method, which is the method for determining the target. That is, the selection unit 133 selects collaborative filtering as the selection logic when a predetermined condition is satisfied.

More specifically, the selection unit 133 includes at least a search method that is a method of determining a transaction target that provides information to the user from a transaction target that is related to the search query entered by the user. Select one of the multiple methods. That is, the selection unit 133 selects history selection as the selection logic when a predetermined condition is satisfied.

For example, the selection unit 133 selects collaborative filtering from the selection logic storage unit 123 when the number or ratio of the behavior information indicating browsing of the transaction target among the behavior information acquired by the transaction target satisfies a predetermined condition. .. Further, for example, when the number or ratio of the action information indicating the browsing of the transaction target among the action information acquired by the transaction target does not satisfy a predetermined condition, the selection unit 133 selects the history from the selection logic storage unit 123. Select.

For example, in the example of FIG. 2, the selection unit 133 calculates a score indicating the possibility that the user selects the transaction target determined by the method for each method based on the behavior, and the calculated method has the highest score. Select. As shown in FIG. 2, the selection unit 133 calculates a score according to the log type for each product selection logic, and selects the selection logic having the highest score.

For example, in the example of FIG. 2, the selection unit 133 calculates a score for each selection logic included in the table TB2. For example, the selection unit 133 calculates that the score corresponding to the collaborative filtering associated with the logic LG1 is 0.5 based on the log included in the table TB1. Further, for example, the selection unit 133 calculates that the score corresponding to the history selection associated with the logic LG2 is 0.2 based on the log included in the table TB1. Further, for example, the selection unit 133 calculates that the score corresponding to the related selection associated with the logic LG 3 is 0.1 based on the log included in the table TB1. Then, the selection unit 133 updates the score stored in the selection logic storage unit 123. Then, the selection unit 133 selects the logic LG1 having the highest score among the plurality of selection logics.

(About estimation unit 134)
The estimation unit 134 estimates the degree of interest of the user in the transaction target related to the behavior indicated by the behavior information for each behavior information. Specifically, the estimation unit 134 uses a model that learns the relationship between the degree of interest of the user in the transaction target related to the behavior indicated by the behavior information and the characteristics of the behavior information for each behavior information. , Estimate the degree of interest of the user.

More specifically, the estimation unit 134 estimates the degree of interest of the user by using a model that learns the relationship between the type of behavior indicated by the behavior information as a feature and the degree of interest of the user. More specifically, the estimation unit 134 uses a model that learns the relationship between whether or not the behavior indicated by the behavior information is browsing of a transaction target related to the behavior and the degree of interest of the user. Estimate the degree of interest in.

More specifically, the estimation unit 134 uses a model that learns the relationship between whether or not the behavior indicated by the behavior information is a web search related to the transaction target and the degree of interest of the user. Estimate the degree of interest. More specifically, the estimation unit 134 determines whether or not the action indicated by the behavior information is a web search of information related to the transaction target, and whether or not the behavior indicates the information of the transaction target is displayed in the search result. The degree of interest of the user is estimated using the model that learned the relationship with the degree of interest of.

More specifically, the estimation unit 134 has a relationship between whether or not the behavior indicated by the behavior information is a web search displaying search results including links to information related to the transaction target and the degree of interest of the user. The degree of interest of the user is estimated using the model that learned the sex.

For example, the estimation unit 134 estimates the CTR of each log using the learning model 122 stored in the storage unit 120. For example, in the example of FIG. 3, the estimation unit 134 estimates the CTR as “0.2” from the information about the log L21 by using the learning model 122. Further, for example, the estimation unit 134 estimates the CTR as “0.002” from the information regarding the log L22 by using the learning model 122. Further, for example, the estimation unit 134 estimates the CTR as “0.0001” from the information regarding the log L23 by using the learning model 122.

Further, for example, the estimation unit 134 estimates the CTR to be "0.01" from the information regarding the log L24 by using the learning model 122. Further, for example, the estimation unit 134 estimates the CTR as “0.05” from the information regarding the log L25 by using the learning model 122.

(About decision unit 135)
The determination unit 135 uses the method selected by the selection unit 133 to determine the transaction target for providing information to the user from the behavior information acquired by the acquisition unit 131. In addition, the determination unit 135 determines a transaction target for providing information to the user based on the behavior information whose degree estimated by the estimation unit 134 satisfies a predetermined condition. Specifically, the determination unit 135 provides information to the user from the behavioral information whose degree estimated by the estimation unit 134 satisfies a predetermined condition by using the method selected by the selection unit 133. Determine the target.

For example, the determination unit 135 determines a product by using a log in which the CTR is equal to or higher than a predetermined threshold value among the CTRs estimated by using the learning model 122 stored in the storage unit 120. For example, in the example of FIG. 3, it is assumed that the logs whose CTR is equal to or higher than a predetermined threshold are the log L10 and the log L20. It is also assumed that collaborative filtering is selected as the selection logic. In this case, the determination unit 135 determines the product to be provided to the target user based on the log in which the CTR is equal to or higher than a predetermined threshold value by using collaborative filtering.

For example, in the example of FIG. 3, the determination unit 135 determines the product MD1 from the information about the log L10 by using selection logic such as collaborative filtering. Further, for example, in the example of FIG. 3, the determination unit 135 determines the product MD5 from the information regarding the log L20 by using selection logic such as collaborative filtering.

(About the provider 136)
The providing unit 136 transmits various information. Specifically, the providing unit 136 provides the user with content including information about the product. For example, the providing unit 136 provides the user U100 with the content CR1 including information on a plurality of products. For example, in the example of FIG. 1, the providing unit 136 has the product A, the price of the product A "1200 yen", the price of the product B and the product B "1500 yen", the price of the product C and the price of the product C " The content CR1 in which "1700 yen" is displayed at a predetermined location is provided to the user U100.

[1-3. Processing procedure (1) Decision processing]
Next, the procedure of the determination process executed by the calculation device 100 according to the first embodiment will be described with reference to FIG. 7. FIG. 7 is a flowchart showing an example of the flow of the determination process executed by the calculation device 100 according to the first embodiment.

As shown in FIG. 7, the acquisition unit 131 acquires the log of the target user (step S101). Then, when the acquisition unit 131 has not acquired the log of the target user (step S101; No), the selection unit 133 waits until the log of the target user is acquired.

On the other hand, when the acquisition unit 131 acquires the log of the target user (step S101; Yes), the selection unit 133 selects a product based on whether or not the log type satisfies a predetermined condition. One of the selection logics is selected (step S102).

For example, in the example of FIG. 2, the calculation device 100 calculates a score indicating the possibility that the user selects the transaction target determined by the method for each method based on the behavior, and the calculated score is the highest method. Select. As shown in FIG. 2, the calculation device 100 calculates a score according to the log type for each product selection logic, and selects the selection logic having the highest score.

For example, in the example of FIG. 2, the selection unit 133 calculates a score for each selection logic included in the table TB2. For example, the selection unit 133 calculates that the score corresponding to the collaborative filtering associated with the logic LG1 is 0.5 based on the log included in the table TB1. Further, for example, the selection unit 133 calculates that the score corresponding to the history selection associated with the logic LG2 is 0.2 based on the log included in the table TB1. Further, for example, the selection unit 133 calculates that the score corresponding to the related selection associated with the logic LG 3 is 0.1 based on the log included in the table TB1. Then, the selection unit 133 updates the score stored in the selection logic storage unit 123. Then, the selection unit 133 selects the logic LG1 having the highest score among the plurality of selection logics.

Then, the determination unit 135 determines the product using the selection logic selected by the selection unit 133 (step S103). For example, the determination unit 135 uses the selection logic with the highest score to determine the product. For example, it is assumed that collaborative filtering is selected as the selection logic. In this case, the determination unit 135 determines the product to be provided to the target user by using collaborative filtering. For example, in the example of FIG. 2, the determination unit 135 determines the product MD1 from the log L10 by using collaborative filtering. Further, for example, in the example of FIG. 2, the calculation device 100 determines the product MD5 from the log L20 by using collaborative filtering.

Then, the providing unit 136 provides the user with the content including the information about the product (step S104). For example, the providing unit 136 provides the user U100 with the content CR1 including information on a plurality of products. For example, in the example of FIG. 1, the providing unit 136 has the product A, the price of the product A "1200 yen", the price of the product B, the price of the product B "1500 yen", the price of the product C, and the price of the product C " The content CR1 in which "1700 yen" is displayed at a predetermined location is provided to the user U100.

[1-4. Processing procedure (2) Log estimation processing]
Next, the procedure of the log estimation process executed by the calculation device 100 according to the first embodiment will be described with reference to FIG. FIG. 8 is a flowchart showing an example of a flow of log estimation processing executed by the calculation device 100 according to the first embodiment.

As shown in FIG. 8, the acquisition unit 131 acquires the user's log (step S201). Then, when the acquisition unit 131 has not acquired the user's log (step S201; No), the generation unit 132 waits until the user's log is acquired.

On the other hand, when the acquisition unit 131 acquires the user's log (step S201; Yes), the generation unit 132 generates a learning model M1 that learns the relationship between the log type and the content of the product corresponding to the log with the CTR. (Step S202). For example, the generation unit 132 generates the learning model M1 by learning the relationship between the input information such as the log type, the date and time, the product, and the category, and the CTR calculated from the number of clicks on the content.

Then, the estimation unit 134 estimates the CTR of each log using the learning model M1 generated by the generation unit 132, and estimates the log whose CTR is equal to or higher than a predetermined threshold value as a useful log (step S203). For example, in the example of FIG. 3, the estimation unit 134 estimates the CTR as “0.2” from the information about the log L21 by using the learning model M1. Further, for example, the estimation unit 134 estimates the CTR to be "0.002" from the information regarding the log L22 by using the learning model M1. Further, for example, the estimation unit 134 estimates the CTR as “0.0001” from the information regarding the log L23 by using the learning model M1.

Further, for example, the estimation unit 134 estimates the CTR to be "0.01" from the information regarding the log L24 by using the learning model M1. Further, for example, the estimation unit 134 estimates the CTR to be "0.05" from the information regarding the log L25 by using the learning model M1. Then, the estimation unit 134 estimates a log whose CTR is equal to or higher than a predetermined threshold value as a useful log among the CTRs estimated using the learning model M1. For example, in the example of FIG. 3, it is assumed that the logs whose CTR is equal to or higher than a predetermined threshold are the log L10 and the log L20. In this case, the estimation unit 134 estimates the log L10 and the log L20 as useful logs.

[1-5. Modification example]
The calculation device 100 described above may be implemented in various different forms other than the above embodiment. Therefore, another embodiment of the calculation device 100 will be described below.

[1-5-1. Selection process (1) Shopping]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 may select a method based on the number or ratio of behavioral information indicating browsing of the transaction target.

For example, it is assumed that the number of times the product included in the log is viewed is "20". Further, it is assumed that collaborative filtering is selected as the selection logic when the number of times the product is viewed is equal to or greater than a predetermined threshold value. Further, when the number of times the product is viewed is smaller than a predetermined threshold value, history selection is selected as the selection logic. Further, it is assumed that the predetermined threshold value is "50". In this case, since the number of times the product included in the log is viewed by the selection unit 133 is "20", which is smaller than a predetermined threshold value, history selection may be selected as the selection logic.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment selects the method based on the number or ratio of the behavioral information indicating the browsing of the transaction target, so that the information that the user is more interested in. Can be provided.

[1-5-2. Selection process (2) Search query]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 may select a method based on the number or ratio of behavioral information indicating a web search related to the transaction target.

For example, it is assumed that the number of searches for products included in the log is "100". Further, it is assumed that collaborative filtering is selected as the selection logic when the number of product searches is equal to or greater than a predetermined threshold value. Further, it is assumed that history selection is selected as the selection logic when the number of product searches is smaller than a predetermined threshold value. Further, it is assumed that the predetermined threshold value is "50". In this case, the selection unit 133 may select collaborative filtering as the selection logic because the number of searches for the products included in the log is "100" and is equal to or greater than a predetermined threshold value.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment selects the method based on the number or ratio of the behavioral information indicating the web search related to the transaction target, so that the user is more strongly interested. Can provide information that draws.

[1-5-3. Selection process (3) Commerce intention]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 performs a method based on the number or ratio of behavioral information indicating a web search in which the information of the transaction target is displayed in the search result, which is a web search of information related to the transaction target. You may choose.

For example, it is assumed that the number of searches related to the search for information related to the product included in the log and the product information is displayed in the search result is "100". Further, it is assumed that collaborative filtering is selected as the selection logic when the number of searches is equal to or greater than a predetermined threshold value. Further, it is assumed that history selection is selected as the selection logic when the number of searches is smaller than a predetermined threshold value. Further, it is assumed that the predetermined threshold value is "50". In this case, the selection unit 133 is a search for information related to the product included in the log, and the number of searches related to the search in which the product information is displayed in the search result is "100", which is equal to or higher than a predetermined threshold value. Therefore, collaborative filtering may be selected as the selection logic.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment is a web search for information related to the transaction target, and is an action information indicating a web search in which the information of the transaction target is displayed in the search result. Since the method is selected based on the number or ratio, it is possible to provide information that is more strongly interesting to the user.

[1-5-4. Selection process (4) Link]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 may select a method based on the number or ratio of behavioral information indicating a web search in which search results including links to information related to the transaction target are displayed.

For example, it is assumed that the number of searches related to the search in which the search result including the link to the information related to the product included in the log is displayed is "100". Further, it is assumed that collaborative filtering is selected as the selection logic when the number of searches is equal to or greater than a predetermined threshold value. Further, it is assumed that history selection is selected as the selection logic when the number of searches is smaller than a predetermined threshold value. Further, it is assumed that the predetermined threshold value is "50". In this case, the selection unit 133 uses the selection logic because the number of searches related to the search in which the search result including the link to the information related to the product included in the log is displayed is "100", which is equal to or more than a predetermined threshold value. Collaborative filtering may be selected.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment is based on the number or ratio of behavioral information indicating a web search in which search results including links to information related to the transaction target are displayed. Since the method is selected, it is possible to provide information that is more strongly interesting to the user.

[1-5-5. Selection process (5) Depending on the domain]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 may select one of a plurality of methods based on the domain corresponding to the action indicated by the action information.

For example, it is assumed that the number of domains indicated by the content including the product included in the log is "100". Further, it is assumed that collaborative filtering is selected as the selection logic when the number of domains is equal to or greater than a predetermined threshold value. Further, it is assumed that history selection is selected as the selection logic when the number of domains is smaller than a predetermined threshold value. Further, it is assumed that the predetermined threshold value is "50". In this case, the selection unit 133 may select collaborative filtering as the selection logic because the number of domains indicated by the content including the product included in the log is "100" and is equal to or more than a predetermined threshold value.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment selects one of a plurality of methods based on the domain corresponding to the action indicated by the action information, so that the user can select one of the methods. It can provide more intense and interesting information.

[1-5-6. Selection process (6) Elapsed time]
In the above embodiment, an example of the selection process in which the selection unit 133 of the calculation device 100 selects the logic LG1 having the highest score among the plurality of selection logics based on the log has been described, but the present invention is not limited to the above selection process. Specifically, the selection unit 133 may select one of a plurality of methods based on the period elapsed since the action indicated by the action information is performed.

For example, it is assumed that the elapsed time elapsed since the content including the product included in the log is viewed is "1 week". Further, it is assumed that collaborative filtering is selected as the selection logic when the elapsed time is smaller than a predetermined threshold value. Further, it is assumed that history selection is selected as the selection logic when the elapsed time is equal to or greater than a predetermined threshold value. Further, it is assumed that the predetermined threshold value is "1 month". In this case, the selection unit 133 selects collaborative filtering as the selection logic because the elapsed time elapsed since the content including the product included in the log is viewed is "1 week", which is smaller than a predetermined threshold value. May be good.

As a result, the selection unit 133 of the calculation device 100 according to the first embodiment selects one of the plurality of methods based on the period elapsed from the action indicated by the action information. , Can provide information that users are more strongly interested in.

[1-5-7. Estimate processing (1) Depending on the domain]
In the above embodiment, an example of the estimation process in which the estimation unit 134 of the calculation device 100 estimates the CTR of each log using the learning model M1 has been described, but the present invention is not limited to the above selection process. Specifically, the estimation unit 134 may estimate the degree of interest of the user by using a model that learns the relationship between the domain corresponding to the behavior indicated by the behavior information and the degree of interest of the user. ..

For example, the learning model M1 is generated by learning the relationship between the log type, the date and time, the product, the category, the input information such as the domain indicated by the content, and the CTR calculated from the number of clicks on the content. It shall be. In this case, the estimation unit 134 may estimate the CTR of each log using the learning model M1.

As a result, the estimation unit 134 of the calculation device 100 according to the first embodiment uses a model that learns the relationship between the behavior indicated by the behavior information and the corresponding domain and the degree of interest of the user. Since the degree of interest is estimated, it is possible to provide information that is more strongly interesting to the user.

[1-5-8. Estimating process (2) Elapsed time]
In the above embodiment, an example of the estimation process in which the estimation unit 134 of the calculation device 100 estimates the CTR of each log using the learning model M1 has been described, but the present invention is not limited to the above selection process. Specifically, the estimation unit 134 estimates the degree of interest of the user by using a model that learns the relationship between the period elapsed since the action indicated by the behavior information is performed and the degree of interest of the user. You may.

For example, the learning model M1 learns the relationship between the log type, the date and time, the product, the category, the input information such as the elapsed time since browsing the content, and the CTR calculated from the number of clicks on the content. By doing so, it is assumed that it has been generated. In this case, the estimation unit 134 may estimate the CTR of each log using the learning model M1.

In the above example, the degree of interest of the user is estimated by using a model in which the estimation unit 134 learns the relationship between the period elapsed since the action indicated by the behavior information is performed and the degree of interest of the user. Although the explanation has been given with an example of the above, the degree of interest of the user may be estimated by using a model learned by considering the log for which the period elapsed since the action indicated by the behavior information is performed as important. For example, it is assumed that a log whose elapsed time is shorter than a predetermined threshold is selected. The log at this time is a log type, a date and time, a product, and a category. Further, it is assumed that the learning model M1 is generated by learning the relationship between the input information, which is a log whose elapsed time is shorter than a predetermined threshold value, and the CTR calculated from the number of clicks on the content. In this case, the estimation unit 134 may estimate the CTR of each log using the learning model M1.

As a result, the estimation unit 134 of the calculation device 100 according to the first embodiment uses a model that learns the relationship between the period elapsed since the action indicated by the action information is performed and the degree of interest of the user. Since the degree of interest of the user is estimated, it is possible to provide information that the user is more strongly interested in.

[1-5-9. Estimating process (3) number of times]
In the above embodiment, an example of the estimation process in which the estimation unit 134 of the calculation device 100 estimates the CTR of each log using the learning model M1 has been described, but the present invention is not limited to the above selection process. Specifically, the estimation unit 134 relates the degree of interest of the user to the transaction target related to the behavior indicated by the behavior information and the number of times the decision unit 135 uses the behavior information when determining the transaction target. The degree of interest of the user may be estimated for each behavioral information using the model learned from. Here, the estimation unit 134 calculates the number of times the action information is used when determining the transaction target by counting the number of times each time the action information is used.

For example, it is assumed that the estimation unit 134 counts the number of times the log is used when the product is determined in advance. In addition, the learning model M1 learns the relationship between the input information such as the log type, the date and time, the product, the category, and the number of times the log is used when determining the product, and the CTR calculated from the number of clicks on the content. By doing so, it is assumed that it has been generated. In this case, the estimation unit 134 may estimate the CTR of each log using the learning model M1.

As a result, the estimation unit 134 of the calculation device 100 according to the first embodiment determines the degree of interest of the user in the transaction target related to the action indicated by the action information, and the action when the decision unit 135 determines the transaction target. Since the degree of interest of the user is estimated for each behavioral information by using the model that has learned the relationship with the number of times the information is used, it is possible to provide information that the user is more strongly interested in.

[1-5-10. Decision processing]
In the above embodiment, an example of the determination process in which the determination unit 135 of the calculation device 100 determines the product to be provided to the target user based on the log in which the CTR is equal to or higher than a predetermined threshold value by using collaborative filtering has been described. , The determination process is not limited to the above. Specifically, the selection unit 133 may select two or more methods from the plurality of methods, and the determination unit 135 may determine the transaction target using each method selected by the selection unit 133. ..

More specifically, the determination unit 135 may determine a different number of transaction targets for each method. Then, the determination unit 135 narrows down the transaction targets to be displayed as information based on the relationships between the transaction targets determined by using various methods.

For example, it is assumed that the products selected by collaborative filtering are product MD1, product MD2, and product MD3. Further, it is assumed that the products selected by the history selection are the product MD1, the product MD3, and the product MD5. In this case, the determination unit 135 determines the product MD1, the product MD2, and the product MD3 from the useful logs estimated based on the learning model M1 by using collaborative filtering. In addition, the determination unit 135 determines the product MD1, the product MD3, and the product MD5 from the useful logs estimated based on the learning model M1 by using the history selection. Then, the determination unit 135 may determine the product MD1 and the product MD3 as products to be provided upward to the user based on the relationship between the products determined by using various methods.

As a result, since the determination unit 135 of the calculation device 100 according to the first embodiment determines the transaction target by using each method, it is possible to provide information that is more strongly interesting to the user. Further, since the determination unit 135 of the calculation device 100 according to the first embodiment determines a different number of transaction targets for each method, it is possible to provide information that is more strongly interesting to the user. Further, the determination unit 135 of the calculation device 100 according to the first embodiment is used because it narrows down the transaction targets to be displayed as information based on the relationships between the transaction targets determined by using various methods. Can provide information that is more strongly interesting to the person.

It should be noted that the number of products to be selected by each method may be arbitrarily set. For example, the higher the score, the more products may be selected. For example, it is assumed that the collaborative filtering score is 0.5, the history selection score is 0.5, and the number of products is 6. In this case, the determination unit 135 uses collaborative filtering to select three products from the useful logs estimated based on the learning model M1, and uses history selection to estimate the usefulness based on the learning model M1. You may decide 3 products from the log. Further, for example, it is assumed that the collaborative filtering score is 0.8, the history selection score is 0.4, and the number of products is 6. In this case, the determination unit 135 uses collaborative filtering to select four products from the useful logs estimated based on the learning model M1, and uses history selection to estimate the usefulness based on the learning model M1. You may decide that there are two products from the log.

[1-5-11. Learning model]
In the above embodiment, an example in which the calculation device 100 selects the selection logic depending on whether or not the log satisfies the condition is described separately from the estimation of the useful log, but the present invention is not limited to the above embodiment. The selection logic may be selected depending on whether the estimated useful log satisfies the condition. For example, the calculation device 100 selects one of the selection logics for selecting products based on whether or not the useful log type estimated using the learning model M1 satisfies a predetermined condition. You may.

Further, the calculation device 100 does not have to select the selection logic. For example, the calculation device 100 may determine a product only by a predetermined selection logic (for example, collaborative filtering) using a useful log. Further, when the selection logic is selected, the calculation device 100 may determine the product based on all the logs other than the useful log. That is, the above-mentioned selection process and estimation process may be performed in any combination or individually.

[1-5-12. CTR]
In the above embodiment, an example of the estimation process in which the calculation device 100 estimates the CTR of each log using the learning model M1 has been described, but it does not have to be limited to the CTR, and any index is estimated as a reaction by the user. You may. For example, the calculation device 100 may estimate the CVR (Conversion Rate) of each log by using the learning model M1.

[2. Second Embodiment]
[2-1. An example of the determination process shown by the calculation device]
Next, the second embodiment will be described. In the first embodiment described above, the calculation device 100 has described an example of a process of providing a user with content including information about a product determined based on a useful log using selection logic. In the second embodiment, the calculation device 200 shows an example of a process of providing a content including information about a product determined based on a useful log using a selection logic to a user who has a high purchase intention.

First, an example of the determination process executed by the calculation device 200 according to the second embodiment will be described with reference to FIG. FIG. 9 is a diagram showing an example of a determination process executed by the calculation device 200 according to the second embodiment. Specifically, the calculation device 200 determines the possibility that the user purchases the transaction target based on whether or not the behavior information indicating the user's behavior includes the behavior information indicating the browsing of the transaction target. Calculate the indicated score.

Since the behavior information referred to here is the same as the content described in the example of the first embodiment, the description thereof will be omitted. Further, since the transaction target referred to here is the same as the content described in the example of the first embodiment, the description thereof will be omitted. In addition, as in the example of the first embodiment, an example will be described below assuming that the transaction target is a product.

As shown in FIG. 9, the calculation system 2 includes a terminal device 10 and a calculation device 200. The terminal device 10 and the calculation device 200 are connected by wire or wireless communication via a predetermined communication network (not shown). The calculation system 2 shown in FIG. 9 may include a plurality of terminal devices 10 and a plurality of calculation devices 200.

Since the terminal device 10 according to the second embodiment has the same configuration as the terminal device 10 according to the first embodiment, the description thereof will be omitted.

The calculation device 200 according to the second embodiment is realized by, for example, a server device or the like. Specifically, the calculation device 200 acquires behavior information indicating the behavior of the user. Then, the calculation device 200 calculates a score indicating the possibility of the user purchasing the transaction target based on whether or not the behavior information includes the behavior information indicating the browsing of the transaction target. In the following, it is assumed that the information about the product is determined in advance based on the useful log using the selection logic.

Hereinafter, an example of the determination process by the calculation device 200 will be described along the flow with reference to FIG.

First, as shown in FIG. 9, the calculation device 200 acquires a log from the user U200 (step S41). Subsequently, the calculation device 200 acquires a log from the user U201 (step S42). Then, the calculation device 200 calculates a purchase score, which is a score indicating the possibility of the user purchasing the product, based on whether or not the log contains behavioral information indicating the browsing of the product (step S43). ). That is, the calculation device 200 calculates a purchase score that reflects each user's willingness to purchase in order to provide different contents including information about the product according to the purchase intention of the user.

Specifically, the calculation device 200 calculates a score indicating the possibility of the user purchasing the transaction target based on whether or not the behavior information indicating the browsing of the transaction target is included. More specifically, the calculation device 200 specifies the browsing frequency of the content related to the transaction target based on the behavior information, and calculates the score of the trading target based on the specified browsing frequency.

For example, in the example of FIG. 9, the purchase score is calculated based on a predetermined formula showing a large value in proportion to the number of times the content related to the user's product is viewed. Further, the purchase score shall indicate a value in the range of 0 to 1. Further, it is assumed that the number of times of viewing the content related to the product of the user U200 is "200". Further, it is assumed that the number of times of viewing the content related to the product of the user U201 is "50". In this case, the calculation device 200 calculates the purchase score of the user U200 as "0.8" based on the number of views "200". Further, the calculation device 200 calculates the purchase score of the user U201 as "0.2" based on the number of views "50".

Subsequently, the calculation device 200 determines whether or not to provide information about the product based on the purchase score (step S44). That is, in order to provide the content including the information about the product according to the purchase intention of the user, the calculation device 200 provides the content including the information about the product based on the purchase score reflecting the purchase intention of each user. Judge whether or not. Specifically, the calculation device 200 determines whether or not to display the information regarding the transaction target on the content of the domain different from the electronic shopping street based on the score. For example, in the example of FIG. 9, it is determined that information about the product is provided when the purchase score is equal to or higher than a predetermined threshold value. It is assumed that the purchase score of the user U200 is "0.8". Further, it is assumed that the purchase score of the user U201 is "0.2". Further, it is assumed that the predetermined threshold value is "0.7". In this case, the calculation device 200 determines that the purchase score of the user U200 is equal to or higher than a predetermined threshold value and therefore provides information about the product. On the other hand, the calculation device 200 determines that the purchase score of the user U201 is smaller than a predetermined threshold value and therefore does not provide information about the product.

Then, the calculation device 200 provides the user U200 with the content CR2 including the information about the product based on the determination result (step S45). For example, the calculation device 200 provides the user with content including information about a product based on a useful log using selection logic based on the determination result. For example, in the example of FIG. 9, the calculation device 200 has the product A and the price of the product A "1200 yen", the price of the product B and the product B "1500 yen", the price of the product C and the price of the product C " The content CR2 in which "1700 yen" is displayed at a predetermined location is provided to the user U200.

Subsequently, the calculation device 200 provides the user U201 with the content CR3 that does not include information about the product based on the determination result (step S46). For example, the calculation device 200 provides the user U201 with content CR3 that does not include information about the product based on a useful log using the selection logic based on the determination result.

In this way, the calculation device 200 according to the second embodiment acquires the behavior information indicating the behavior of the user. Then, the calculation device 200 calculates a score indicating the possibility of the user purchasing the transaction target based on whether or not the behavior information includes the behavior information indicating the browsing of the transaction target. As a result, the calculation device 200 according to the second embodiment calculates a score indicating the possibility that the user purchases the transaction target based on whether or not the behavior information indicating the browsing of the transaction target is included. Therefore, it is possible to provide information that is more strongly interesting to the user.

This point will be described. Explaining with reference to the example of FIG. 9, the calculation device 200 determines whether or not to provide the information about the product to the user based on the purchase score. As a result, the calculation device 200 is expected to have a high possibility that a user having a high purchase score will purchase the product, and thus provides content including information about the product. On the other hand, the calculation device 200 provides content that does not include information about the product because it is predicted that a user with a low purchase score is unlikely to purchase the product. In this way, the calculation device 200 can provide and separate the content according to the purchase intention of the user, so that the appealing effect of the content can be enhanced. Further, since the calculation device 200 provides the content including the information about the product to the user who is expected to have a high purchase intention, it is possible to efficiently provide the information that the user is more strongly interested in. .. Therefore, the calculation device 100 can efficiently provide information that is more strongly and interesting to the user, and thus can provide a better service.

[2-2. Configuration of calculation device]
Next, the configuration of the calculation device 200 according to the second embodiment will be described with reference to FIG. FIG. 10 is a diagram showing a configuration example of the calculation device 200 according to the second embodiment. As shown in FIG. 10, the calculation device 200 includes a communication unit 110, a storage unit 120, and a control unit 230.

(About control unit 230)
The control unit 230 is a controller, and for example, various programs (corresponding to an example of a calculation program) stored in a storage device inside the calculation device 200 are executed by a CPU, MPU, or the like using the RAM as a work area. It will be realized. Further, the control unit 230 is a controller, and is realized by, for example, an integrated circuit such as an ASIC or FPGA.

As shown in FIG. 10, the control unit 230 includes an acquisition unit 131, a generation unit 132, a selection unit 133, an estimation unit 134, a determination unit 135, a calculation unit 231 and a determination unit 232, and a provision unit 136. To realize or execute the functions and actions of information processing described below. The internal configuration of the control unit 230 is not limited to the configuration shown in FIG. 10, and may be any other configuration as long as it performs information processing described later. Further, the connection relationship of each processing unit included in the control unit 230 is not limited to the connection relationship shown in FIG. 10, and may be another connection relationship.

(About calculation unit 231)
The calculation unit 231 calculates a score indicating the possibility that the user purchases the transaction target based on whether or not the behavior information acquired by the acquisition unit 131 includes the behavior information indicating the browsing of the transaction target. To do. Specifically, the calculation unit 231 specifies the browsing frequency of the content related to the transaction target based on the behavior information, and calculates the score of the trading target based on the specified browsing frequency.

For example, in the example of FIG. 9, the purchase score is calculated based on a predetermined formula showing a large value in proportion to the number of times the content related to the user's product is viewed. Further, the purchase score shall indicate a value in the range of 0 to 1. Further, it is assumed that the number of times of viewing the content related to the product of the user U200 is "200". Further, it is assumed that the number of times of viewing the content related to the product of the user U201 is "50". In this case, the calculation device 200 calculates the purchase score of the user U200 as "0.8" based on the number of views "200". Further, the calculation device 200 calculates the purchase score of the user U201 as "0.2" based on the number of views "50".

(About judgment unit 232)
The determination unit 232 determines whether or not to display the information regarding the transaction target on the content of the domain different from the electronic shopping street, based on the score calculated by the calculation unit 231. For example, in the example of FIG. 9, it is determined that information about the product is provided when the purchase score is equal to or higher than a predetermined threshold value. It is assumed that the purchase score of the user U200 is "0.8". Further, it is assumed that the purchase score of the user U201 is "0.2". Further, it is assumed that the predetermined threshold value is "0.7". In this case, the calculation device 200 determines that the purchase score of the user U200 is equal to or higher than a predetermined threshold value and therefore provides information about the product. On the other hand, the calculation device 200 determines that the purchase score of the user U201 is smaller than a predetermined threshold value and therefore does not provide information about the product.

[2-3. Processing procedure]
Next, the procedure of the determination process executed by the calculation device 200 according to the second embodiment will be described with reference to FIG. FIG. 11 is a flowchart showing an example of the flow of the determination process executed by the calculation device according to the second embodiment.

As shown in FIG. 11, the acquisition unit 131 acquires the log of the target user (step S301). Then, when the acquisition unit 131 has not acquired the log of the target user (step S301; No), the calculation unit 231 waits until the log of the target user is acquired.

On the other hand, when the acquisition unit 131 acquires the log of the target user (step S301; Yes), the calculation unit 231 purchases the product based on whether or not the log contains behavioral information indicating browsing of the product. The score is calculated (step S302).

For example, in the example of FIG. 9, the purchase score is calculated based on a predetermined formula showing a large value in proportion to the number of times the content related to the user's product is viewed. Further, the purchase score shall indicate a value in the range of 0 to 1. Further, it is assumed that the number of times of viewing the content related to the product of the user U200 is "200". Further, it is assumed that the number of times of viewing the content related to the product of the user U201 is "50". In this case, the calculation device 200 calculates the purchase score of the user U200 as "0.8" based on the number of views "200". Further, the calculation device 200 calculates the purchase score of the user U201 as "0.2" based on the number of views "50".

Then, the determination unit 232 determines whether or not to provide information about the product based on the purchase score calculated by the calculation unit 231 (step S303). For example, in the example of FIG. 9, it is determined that information about the product is provided when the purchase score is equal to or higher than a predetermined threshold value. It is assumed that the purchase score of the user U200 is "0.8". Further, it is assumed that the purchase score of the user U201 is "0.2". Further, it is assumed that the predetermined threshold value is "0.7". In this case, the calculation device 200 determines that the purchase score of the user U200 is equal to or higher than a predetermined threshold value and therefore provides information about the product. On the other hand, the calculation device 200 determines that the purchase score of the user U201 is smaller than a predetermined threshold value and therefore does not provide information about the product.

Then, when it is determined by the determination unit 232 that the information about the product is provided by the providing unit 136 (step S303; Yes), the providing unit 136 provides the user with the content including the information about the product (step S304). For example, the providing unit 136 provides the user with content including information about the product based on a useful log by using the selection logic based on the determination result determined by the determination unit 232. For example, in the example of FIG. 9, the providing unit 136 has the product A, the price of the product A "1200 yen", the price of the product B, the price of the product B "1500 yen", the price of the product C, and the price of the product C " The content CR2 in which "1700 yen" is displayed at a predetermined location is provided to the user U200.

On the other hand, when the determination unit 232 determines that the information about the product is not provided (step S303; No), the providing unit 136 provides the user with the content that does not include the information about the product (step S305). For example, the calculation device 200 provides the user U201 with content CR3 that does not include information about the product based on a useful log using the selection logic based on the determination result.

[2-4. Modification example]
The calculation device 200 described above may be implemented in various different forms other than the above embodiment. Therefore, in the following, another embodiment of the calculation device 200 will be described.

[2-4-1. Calculation processing (1) Information on solvency]
In the above embodiment, an example of the calculation process in which the calculation unit 231 of the calculation device 200 calculates the purchase score, which is the score indicating the possibility of the user purchasing the product, has been described based on the log. Not limited. Specifically, the acquisition unit 131 may acquire the ability information indicating the purchasing ability of the transaction target by the user, and the calculation unit 231 may calculate the score based on the behavior information and the ability information. The ability information referred to here is, for example, the credit of the user, the purchase amount of the user, the credit card usage history of the user, the rank of the credit card of the user, and the like.

For example, the purchase score shall be calculated based on a predetermined formula showing a large value in proportion to the number of views of the content related to the user's product and the credit score. Further, the purchase score shall indicate a value in the range of 0 to 1. Further, it is assumed that the number of times the user U200 has viewed the product is "200". Further, it is assumed that the number of times the user U201 has viewed the product is "50". Further, it is assumed that the credit score indicating the degree of credit of the user U200 is "0.9". Further, it is assumed that the credit score indicating the degree of credit of the user U201 is "0.5". In this case, the calculation device 200 calculates the purchase score of the user U200 as "0.9" based on the number of views "200" and the credit score "0.9". Further, the calculation device 200 may calculate the purchase score of the user U201 as "0.3" based on the number of views "50" and the credit score "0.5".

As a result, the calculation unit 231 of the calculation device 200 according to the second embodiment calculates the score based on the behavior information and the ability information, so that the user can provide information that is more strongly interesting. it can.

[2-4-2. Calculation process (2) Viewing video]
In the above embodiment, an example of the calculation process in which the calculation unit 231 of the calculation device 200 calculates the purchase score, which is the score indicating the possibility of the user purchasing the product, has been described based on the log. Not limited. Specifically, the calculation unit 231 may calculate the score based on whether or not the action information further includes the action information indicating the viewing of the moving image regarding the transaction target.

For example, the purchase score shall be calculated based on a predetermined formula showing a large value in proportion to the number of times the user has viewed the content related to the product and the number of times the video related to the product has been viewed. Further, the purchase score shall indicate a value in the range of 0 to 1. Further, it is assumed that the number of times the user U200 has viewed the product is "200". Further, it is assumed that the number of times the user U201 has viewed the product is "50". Further, it is assumed that the number of times of viewing the video related to the product of the user U200 is "100". Further, it is assumed that the number of times of viewing the video related to the product of the user U201 is "50". In this case, the calculation device 200 calculates the purchase score of the user U200 as "0.8" based on the number of views "200" and the number of views of the moving image "100". Further, the calculation device 200 may calculate the purchase score of the user U201 as "0.3" based on the number of views "50" and the number of views "50" of the moving image.

As a result, the calculation unit 231 of the calculation device 200 according to the second embodiment calculates the score based on whether or not the action information further includes the action information indicating the viewing of the moving image regarding the transaction target. Therefore, it is possible to provide information that is more strongly interesting to the user.

[2-4-3. Offer processing (1) When purchasing a product]
In the above embodiment, an example of the provision process in which the provision unit 136 of the calculation device 200 provides the user with the content including the information about the product has been described, but the present invention is not limited to the selection process. Specifically, the calculation unit 231 determines whether or not the user has purchased the transaction target based on the behavior information, and if so, determines a score indicating the possibility of purchasing the transaction target. Make it a value. Then, the determination unit 232 determines that the information regarding the transaction target is not displayed on the content of the domain different from the electronic shopping street, based on the predetermined value calculated by the calculation unit 231. Then, the providing unit 136 may provide the user with the content that does not include the information about the product.

For example, it is assumed that the predetermined value is "0.0". Further, when the purchase score is equal to or higher than a predetermined threshold value, it is determined that the information about the product is provided. Further, it is assumed that the predetermined threshold value is "0.7". In this case, the calculation unit 231 sets the purchase score to a predetermined value "0.0" when the user purchases the product. Then, the determination unit 232 determines that the information about the product is not provided because the purchase score of the user is smaller than the predetermined threshold value. Then, the providing unit 136 may provide the user with the content that does not include the information about the product based on the determination result.

As a result, the providing unit 136 of the calculation device 200 according to the second embodiment determines whether or not the user has purchased the transaction target based on the behavior information, and if so, purchases the transaction target. Since the score indicating the possibility of doing so is set to a predetermined value, it is possible to provide information that is more strongly interesting to the user.

[2-4-4. Offer processing (2) Coupon]
In the above embodiment, an example of the provision process in which the provision unit 136 of the calculation device 200 provides the user with the content including the information about the product has been described, but the present invention is not limited to the selection process. Specifically, when the score calculated by the calculation unit 231 satisfies a predetermined condition, the provision unit 136 provides the user with the profit of processing for the purchase of the transaction target related to the behavior indicated by the behavior information. May be good. The processing profit referred to here is, for example, a coupon that can be used when purchasing a product.

For example, when the purchase score is equal to or higher than a predetermined threshold value, a coupon that can be used when purchasing a product shall be provided. In addition, it is assumed that the purchase score of the user is "0.9". Further, it is assumed that the predetermined threshold value is "0.7". In this case, since the purchase score of the user is "0.9" and is equal to or higher than a predetermined threshold value, the providing unit 136 provides the user with information about the product and content including a coupon that can be used when purchasing the product. You may.

As a result, when the score satisfies a predetermined condition, the providing unit 136 of the calculation device 200 according to the second embodiment provides the user with the profit of processing for the purchase of the transaction target related to the action indicated by the action information. Therefore, it is possible to provide information that is more strongly interesting to the user.

[2-4-5. Offer processing (3) Discount rate]
In the above embodiment, an example of the provision process in which the provision unit 136 of the calculation device 200 provides the user with the content including the information about the product has been described, but the present invention is not limited to the selection process. Specifically, the providing unit 136 may provide the user with a benefit having different contents depending on the value of the score. The profits having different contents referred to here are, for example, discount rates related to products.

For example, the discount rate shall be calculated based on a predetermined formula showing a value larger in proportion to the purchase score. Further, the discount rate shall indicate a value in the range of 0 to 0.5. In addition, it is assumed that the purchase score of the user is "0.9". In this case, the provision unit 136 calculates the discount rate that can be used by the user as "0.5" based on the purchase score "0.9". Then, the providing unit 136 may provide the user with the content including the product and the amount obtained by subtracting the discount rate "0.5" from the price.

As a result, the providing unit 136 of the calculation device 200 according to the second embodiment provides the user with profits having different contents depending on the value of the score, so that the user can provide information that is more strongly interesting. Can be done.

[3. Hardware configuration]
Further, the terminal device 10 and the calculation device 100 or 200 according to the above-described embodiment are realized by, for example, a computer 1000 having a configuration as shown in FIG. Hereinafter, the calculation device 100 will be described as an example. FIG. 12 is a hardware configuration diagram showing an example of a computer 1000 that realizes the functions of the calculation device 100. The computer 1000 includes a CPU 1100, a RAM 1200, a ROM (Read Only Memory) 1300, an HDD (Hard Disk Drive) 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F). ) Has 1700.

The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400, and controls each part. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started, a program that depends on the hardware of the computer 1000, and the like.

The HDD 1400 stores a program executed by the CPU 1100, data used by such a program, and the like. The communication interface 1500 receives data from another device via the network N and sends it to the CPU 1100, and the CPU 1100 transmits the data generated by the CPU 1100 via the network N to the other device.

The CPU 1100 controls an output device such as a display or a printer, and an input device such as a keyboard or a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the data generated via the input / output interface 1600 to the output device.

The media interface 1700 reads a program or data stored in the recording medium 1800 and provides the program or data to the CPU 1100 via the RAM 1200. The CPU 1100 loads the program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable Disk), a magneto-optical recording medium such as MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. And so on.

For example, when the computer 1000 functions as the calculation device 100 according to the embodiment, the CPU 1100 of the computer 1000 realizes the function of the control unit 130 by executing the program loaded on the RAM 1200. Further, the data in the storage unit 120 is stored in the HDD 1400. The CPU 1100 of the computer 1000 reads and executes these programs from the recording medium 1800, but as another example, these programs may be acquired from another device via the network N.

[4. Others]
Further, among the processes described in the above-described embodiments and modifications, all or part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed. It is also possible to automatically perform all or part of the processed processing by a known method. In addition, the processing procedure, specific name, and information including various data and parameters shown in the above document and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

Further, each component of each of the illustrated devices is a functional concept, and does not necessarily have to be physically configured as shown in the figure. That is, the specific form of distribution / integration of each device is not limited to the one shown in the figure, and all or part of the device is functionally or physically distributed / physically in arbitrary units according to various loads and usage conditions. It can be integrated and configured.

In addition, the above-described embodiments and modifications can be appropriately combined as long as the processing contents do not contradict each other.

Further, the above-mentioned "section, module, unit" can be read as "means" or "circuit". For example, the calculation unit can be read as a calculation means or a calculation circuit.

[5. effect〕
As described above, the calculation device 100 according to the embodiment includes the acquisition unit 131 and the calculation unit 231. The acquisition unit 131 acquires behavior information indicating the behavior of the user. The calculation unit 231 calculates a score indicating the possibility that the user purchases the transaction target based on whether or not the behavior information acquired by the acquisition unit 131 includes the behavior information indicating the browsing of the transaction target. To do.

As a result, the calculation device 100 according to the embodiment indicates the possibility that the user purchases the transaction target based on whether or not the acquired behavior information includes the behavior information indicating the browsing of the transaction target. Since the score is calculated, it is possible to provide information that is more strongly interesting to the user.

Further, in the calculation device 100 according to the embodiment, a determination unit that determines whether or not to display information about the transaction target on the content of a domain different from the electronic shopping street based on the score calculated by the calculation unit 231. It has 232.

As a result, the calculation device 100 according to the embodiment determines whether or not to display the information regarding the transaction target on the content of the domain different from the electronic shopping street based on the calculated score, so that the user can determine. It can provide more intense and interesting information.

Further, in the calculation device 100 according to the embodiment, when the score calculated by the calculation unit 231 satisfies a predetermined condition, the user is provided with the profit of processing for the purchase of the transaction target related to the action indicated by the action information. It has a providing unit 136.

As a result, when the calculated score satisfies a predetermined condition, the calculation device 100 according to the embodiment is used to provide the user with the profit of processing for the purchase of the transaction target related to the action indicated by the action information. Can provide information that is more strongly interesting to the person.

Further, in the calculation device 100 according to the embodiment, the providing unit 136 provides the user with a benefit having different contents depending on the value of the score.

As a result, the calculation device 100 according to the embodiment provides the user with profits having different contents depending on the value of the score, so that the user can provide information that is more strongly interesting.

Further, in the calculation device 100 according to the embodiment, the calculation unit 231 determines whether or not the user has purchased the transaction target based on the behavior information, and if so, there is a possibility of purchasing the transaction target. Set the score indicating to a predetermined value.

As a result, the calculation device 100 according to the embodiment determines whether or not the user has purchased the transaction target based on the behavior information, and if so, gives a score indicating the possibility of purchasing the transaction target. Since the value is set to a predetermined value, it is possible to provide information that is more strongly interesting to the user.

Further, in the calculation device 100 according to the embodiment, the acquisition unit 131 acquires the ability information indicating the purchasing ability of the transaction target by the user, and the calculation unit 231 calculates the score based on the behavior information and the ability information. calculate.

As a result, the calculation device 100 according to the embodiment calculates the score based on the behavior information and the ability information, so that the user can provide information that is more strongly interesting.

Further, in the calculation device 100 according to the embodiment, the calculation unit 231 calculates the score based on whether or not the action information further includes the action information indicating the viewing of the moving image regarding the transaction target.

As a result, the calculation device 100 according to the embodiment calculates the score based on whether or not the behavior information further includes the behavior information indicating the viewing of the moving image regarding the transaction target, so that the user is stronger. It can provide interesting information.

Further, in the calculation device 100 according to the embodiment, the calculation unit 231 specifies the browsing frequency of the content related to the transaction target based on the behavior information, and calculates the score of the transaction target based on the specified browsing frequency.

As a result, the calculation device 100 according to the embodiment specifies the browsing frequency of the content related to the transaction target based on the behavior information, and calculates the score of the trading target based on the specified browsing frequency. It can provide strongly interesting information.

Although some of the embodiments of the present application have been described in detail with reference to the drawings, these are examples, and various modifications are made based on the knowledge of those skilled in the art, including the embodiments described in the disclosure column of the invention. It is possible to carry out the present invention in other improved forms.

1 Calculation system 2 Calculation system 10 Terminal device 100 Calculation device 110 Communication unit 120 Storage unit 121 Log information storage unit 122 Learning model 123 Selection logic storage unit 130 Control unit 131 Acquisition unit 132 Generation unit 133 Selection unit 134 Estimation unit 135 Decision unit 136 Providing unit 200 Calculation unit 230 Control unit 231 Calculation unit 232 Judgment unit

Claims (10)

An acquisition unit that acquires behavior information that indicates the behavior of the user,
A calculation unit that calculates a score indicating the possibility that the user will purchase the transaction target based on whether or not the behavior information acquired by the acquisition unit includes behavior information indicating browsing of the transaction target. When,
A calculation device characterized by having. Claim 1 is characterized in that it has a determination unit for determining whether or not to display information on the transaction target on the content of a domain different from the electronic shopping district based on the score calculated by the calculation unit. The calculation device described in. When the score calculated by the calculation unit satisfies a predetermined condition, it is characterized by having a provision unit that provides the user with the profit of processing for the purchase of the transaction target related to the behavior indicated by the behavior information. The calculation device according to claim 1 or 2. The providing part
Providing benefits to the user with different contents depending on the value of the score.
The calculation device according to claim 3, wherein the calculation device is characterized in that. The calculation unit
Based on the behavior information, it is determined whether or not the user has purchased the transaction target, and if so, a score indicating the possibility of purchasing the transaction target is set to a predetermined value.
The calculation device according to any one of claims 1 to 4, wherein the calculation device is characterized in that. The acquisition unit
Acquire ability information indicating the purchasing ability of the transaction target by the user,
The calculation unit
The score is calculated based on the behavior information and the ability information.
The calculation device according to any one of claims 1 to 5, wherein the calculation device is characterized in that. The calculation unit
The score is calculated based on whether or not the behavior information further includes behavior information indicating viewing of a moving image relating to the transaction target.
The calculation device according to any one of claims 1 to 6, characterized in that. The calculation unit
Based on the behavior information, the browsing frequency of the content related to the trading target is specified, and the score of the trading target is calculated based on the specified browsing frequency.
The calculation device according to any one of claims 1 to 7, wherein the calculation device is characterized in that. It is a calculation method executed by the calculation device.
The acquisition process to acquire behavior information showing the behavior of the user,
A calculation step of calculating a score indicating the possibility that the user purchases the transaction target based on whether or not the behavior information acquired by the acquisition process includes behavior information indicating browsing of the transaction target. A calculation method characterized by including and. Acquisition procedure to acquire behavior information showing user behavior,
A calculation procedure for calculating a score indicating the possibility of the user purchasing the transaction target based on whether or not the behavior information acquired by the acquisition procedure includes behavior information indicating browsing of the transaction target. A calculation program characterized by having and.

Images