JP2022027292A - Analysis device, analysis method, and analysis program - Google Patents

Abstract

To enable effective use of history data on usage of (payment for) products and services for usage promotion.SOLUTION: An analysis device 10 of the present invention comprises an acquisition unit and a sorting unit. The acquisition unit acquires histories of usage of given targets by users. The sorting unit refers to a usage history of each user to sort users or targets into groups of similar users or similar targets using appearance frequencies of histories corresponding to pairs of a user and a target used by the user, and reciprocals of appearance frequencies of the targets used by all the users.SELECTED DRAWING: Figure 1

Description

The present invention relates to an analyzer, an analytical method and an analytical program.

In recent years, there are expectations for technology that analyzes historical data of payments for products and services using smartphones on a store-by-store basis and effectively utilizes them for sales promotion.

Japanese Unexamined Patent Publication No. 2014-194710

However, with the prior art, it is difficult to analyze the payment history data for each store because the number of stores is huge and the payment history data is large. Therefore, it was difficult to utilize the payment history data for sales promotion.

The present application has been made in view of the above, and an object thereof is to provide an analysis device, an analysis method, and an analysis program capable of effectively using historical data of use of products, services, etc. for promotion of use. And.

The analyzer according to the present application has an acquisition unit that acquires the usage history of a predetermined target by the user, and the appearance frequency of the history corresponding to the pair of the user and the target used by the user based on the usage history by each user. It is characterized by having a classification unit for classifying into a user group or a target group that are similar to each other by using the reciprocal of the appearance frequency of the target used by all users.

According to one aspect of the embodiment, there is an effect that the history data of the use of products, services, etc. can be effectively used for promotion of use.

FIG. 1 is a diagram showing an example of a classification process executed by the analytical measures according to the embodiment. FIG. 2 is a diagram showing an example of a classification process executed by the analytical measures according to the embodiment. FIG. 3 is a diagram showing a configuration example of the analyzer according to the embodiment. FIG. 4 is a diagram showing an example of user database information according to the embodiment. FIG. 5 is a diagram showing an example of payment history database information according to the embodiment. FIG. 6 is a diagram showing an example of a store database according to an embodiment. FIG. 7 is a sequence diagram showing a processing procedure executed by the analyzer according to the embodiment. FIG. 8 is a hardware configuration diagram showing an example of a computer that realizes the functions of the analyzer according to the embodiment.

Hereinafter, an analyzer, an analysis method, and an embodiment for carrying out an analysis program (hereinafter referred to as “embodiment”) according to the present application will be described in detail with reference to the drawings. It should be noted that this embodiment does not limit the analyzer, analysis method and analysis program according to the present application. Further, in each of the following embodiments, the same parts are designated by the same reference numerals, and duplicate explanations are omitted.

(Embodiment)
[1-1. About payment using user terminal]
First, an example of payment using a user terminal (electronic payment) will be described. In the following description, the user U is a user terminal using a two-dimensional code (QR code (registered trademark)) arranged in the store and indicating the store identification information for identifying the store. An example of performing payment using the above will be described, but the embodiment is not limited to this. The example of payment described below can be applied even when an arbitrary user makes a payment at an arbitrary store using an arbitrary user terminal. Further, the store identification information C1 may be not only a QR code (registered trademark) but also a barcode, a predetermined mark, a number, or the like.

For example, when a user makes a payment associated with the use or purchase of a payment target (transaction target) such as various products or services at a store, the user uses a payment application (hereinafter referred to as “hereinafter”) pre-installed on the user terminal. It may be simply described as "payment application"). Then, the user captures the store identification information installed in the store via the payment application. In such a case, the user terminal displays a screen for inputting the price to be settled, and accepts the input of the payment amount from the user or the store clerk. Then, the user terminal uses the user identification information for identifying the user, the store identification information (or the information indicated by the store identification information, that is, the information indicating the store (for example, the store ID)), and the payment amount. The indicated payment information is sent to the payment server.

In such a case, the payment server transfers the electronic money of the amount indicated by the payment amount from the user's account indicated by the user identification information to the store account indicated by the store identification information. Then, the payment server sends a notification to the user terminal that the payment has been completed. In such a case, the user terminal notifies that the payment has been made by electronic money by outputting a screen indicating that the payment has been completed and a predetermined voice.

It should be noted that the payment using the user terminal is not limited to the above-mentioned processing. For example, the payment using the user terminal may be made using the store terminal installed in the store. For example, the user terminal displays user identification information for identifying the user on the screen. In such a case, the store terminal installed in the store reads the user identification information displayed on the user terminal, and the user identification information (or the information indicated by the user identification information, that is, the information indicating the user). (For example, the user ID)), the payment amount, and the payment information indicating the information for identifying the store are transmitted to the payment server. In such a case, the payment server transfers the electronic money of the amount indicated by the payment amount from the user's account indicated by the user identification information to the store's account, and transfers the electronic money of the amount indicated by the payment amount to the store terminal or the user terminal of the store. You may notify that the payment has been made by outputting a screen indicating that the payment has been completed or a predetermined voice.

In addition, payment using a user terminal is not limited to the process of transferring electronic money from an account charged with electronic money by user U to a store account, for example, using a credit card registered in advance by the user. It may be the payment that was made. In such a case, for example, the user terminal may transfer the electronic money of the payment amount to the store account and charge the payment amount to the management company of the user's credit card.

[1-2. Outline of the embodiment]
Next, the outline of the analyzer according to the embodiment will be described with reference to FIGS. 1 and 2. 1 and 2 are diagrams showing an example of a classification process executed by the analyzer according to the embodiment. In the embodiment, users who have similar tendencies of stores to use are considered to have similar potential tendencies such as region, democratic club, lifestyle, and hobbies. Therefore, as shown in FIG. 1, the analyzer 10 according to the embodiment uses the history data of payment using a smartphone to a group of users (hereinafter referred to as a user cluster) having similar tendencies of stores to use. Classify. Similarly, the analyzer 10 is classified into a group of stores (hereinafter referred to as a store cluster) having similar tendencies of users.

Then, the analysis device 10 identifies a user or a store of each cluster, and outputs the specific result to, for example, an analysis device 20 used by an analyst or a service developer. For example, the analyst can use the output specific result to recommend the users of the user cluster to which the target seed (reference) user belongs to use the store used by the seed user. .. Alternatively, the analyst can recommend advertisements to users of the user cluster who use other stores belonging to the same store cluster for each store. In this way, by using the processing result of the analyzer 10, it is possible to efficiently promote the sales of the store.

Specifically, the analyzer 10 classifies the payment history data into user clusters by using the degree of bias of users in each store as a feature amount. For example, the analyzer 10 uses the appearance frequency represented by the number of times each user has used each store and the reciprocal of the total number of users who have used each store (hereinafter referred to as the reciprocal user frequency). TFIDF (Term Frequency Inverse Document Frequency) is calculated as a score indicating the degree of bias of users in the store.

For example, in the example shown in FIG. 2, the appearance frequency of the store 1 by the user 1 is 3. Further, the total number of users in the store 1 is 3, and the frequency of reverse users is 1/3. In this case, TFIDF has a value according to 3 × (1/3) = 1. Similarly, the frequency of appearance of the store 3 by the user 2 is 5, the total number of users of the store 3 is 1, and the frequency of reverse users is 1/1. In this case, TFIDF has a value according to 5 × (1/1) = 5.

The TFIDF calculation method is not limited to the above, and is another calculation method in which words in a document are weighted using two indexes of word appearance frequency (TF) and inverse document frequency (IDF). You may.

Further, the analyzer 10 classifies users showing the same feature amount into the same user cluster, using the combination of scores of each user for each store as the feature amount. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each user to the seed user, thereby extracting a user cluster consisting of users similar to the seed user.

Specifically, the analyzer 10 trains a logistic regression model (hereinafter, may be simply referred to as a “model”) using the features of the seed user, and uses the trained model to train the seed user. Calculate the similarity probability of each user other than.

For example, when the score of each store used by one or more seed users is used as a feature, a model in which the weight for each feature is learned so that the range of values indicating the seed user (for example, 1) is output. learn. Then, the score of each store used by each user is input to the learned model. In addition, the probability of similarity between each user and the seed user is calculated based on the value output by the model. For example, the analyzer 10 may have a similarity probability based on the ratio of the value output by the model to the value indicating the seed user.

In addition to the logistic regression model, the model may be realized by a perceptron or various neural networks.

Then, the analyzer 10 classifies users having a similarity probability of a predetermined threshold value or more into the same user cluster as the seed user. For example, the analyzer 10 classifies the history data of all payments for each user cluster by using a clustering method such as K-means. In this way, since the classification can be performed without considering the attributes of each user and each store, it is possible to easily classify the history data of large-scale payments.

Furthermore, considering the usage mode of the store by each user, it is difficult to think of a usage mode in which each user uses all the stores in each region, and the store is selected according to the user's attributes (for example, gender, residential area, etc.). It is thought that bias will occur. As a result, the payment history data can be regarded as a kind of matrix including the fact that the user is not using a certain store, that is, a sparse matrix. When the score based on the user frequency and the reverse user frequency is calculated for the data of such a sparse matrix based on the TFIDF method, the term in which the weight other than "0" can be set in the logistic regression model is , Can be limited to the stores and corresponding sections that are characteristic of a group of seed users. As a result, the analyzer 10 can accurately estimate a user similar to the seed user based on a store cluster that can be a feature of a certain seed user group.

The users classified into the same user cluster as the seed users here are users who have similar stores to the seed users but are different from the seed users, which correspond to so-called false positives, or users who have not been observed. .. For example, a false-positive user with the attribute "housewife" is a user whose attribute is not "housewife" but shows the same tendency as a housewife such as a child who helps with housework or a single man who likes cooking. Further, the unobserved user is, for example, a housewife who is not observed as data. As a result, for example, when the attribute of the seed user is "housewife", not only the user whose attribute is "housewife" but also the attribute is not "housewife" or unknown in the same user cluster as the seed user. However, it is possible to include other users whose behaviors are similar to those of users whose attributes are "housewives".

Further, in the above-mentioned process, the store group used by the seed user is set as a store cluster, and the user who uses the same store cluster is set as a similar user of the seed user, and the user cluster is extracted. However, the extraction process of the store cluster is not limited to this, and the store cluster is defined as a group of users who use the seed store as a user cluster, and a store used by users of the same user cluster as the seed store as a store similar to the seed store. It may be extracted.

For example, the analyzer 10 uses the frequency of appearance of each store by each user and the reciprocal of the total number of stores used by each user (hereinafter referred to as the reciprocal store frequency) to determine the degree of bias of the stores used by each user. TFIDF is calculated as the representative score. Then, the analyzer 10 classifies the stores showing the same feature amount into the same store cluster, using the combination of the scores of each user in each store as the feature amount. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each store to the seed store, thereby extracting a store cluster consisting of stores similar to the seed store.

Specifically, the analyzer 10 learns a logistic regression model using the feature quantities of the seed stores, and calculates the similarity probability of each store other than the seed stores using the learned model. Then, the analyzer 10 classifies stores having a similarity probability of a predetermined threshold value or more into the same store cluster as the seed store. For example, the analyzer 10 classifies the history data of all payments for each store cluster by using a clustering method such as K-means.

The analyzer 10 may classify users whose number of stores used in common with the seed user is greater than a predetermined threshold into the same user cluster as the seed user, based on the history of the stores used by each user. .. Similarly, the analyzer 10 may classify a store in which the number of users common to the seed store is larger than a predetermined threshold into the same store cluster as the seed store.

The analyzer 10 identifies the users included in each user cluster classified as described above or the stores included in each store cluster by the user ID that identifies the user included in the payment history data or the store ID that identifies the store. can do. Further, the analyzer 10 holds a user database that manages user attribute information and a store database that manages store attribute information. Therefore, the analyzer 10 can extract the attribute information of the user of each specified user cluster and the attribute information of the store of each specified store cluster. Therefore, as a result of the analysis process, the analysis device 10 outputs the users included in each classified user cluster or the stores included in each store cluster and the attribute information of each to, for example, the analysis device 20.

Then, the analysis device 20 can propose a sales promotion measure to each store by using the result output from the analysis device 10. For example, a user who is known to have the attribute "housewife" can be a seed user, and a user with another attribute in the same user cluster as the seed user can be a new user of the store used by the user with the attribute "housewife". Judgment makes it possible to expand the audience. Therefore, targeting from each store to these users is effective.

Alternatively, the user who actually used the delivery store is set as the seed user, and it is determined that other users included in the same user cluster as the seed user can become the delivery store user in the future, and the audience can be expanded. Therefore, targeting from each store to these users is effective.

In addition, it is possible to clarify the trade area of a store by analyzing the attributes of the stores in the same store cluster. For example, when the address of a store in the same store cluster is in a single area, it is assumed that the store is a community-based store such as a supermarket or a private store in a residential area. On the other hand, when it spans multiple cities, towns and villages, it is assumed that it is a cross-border store that is used in a wide area across a wide area such as convenience stores and chain stores along the main road.

Furthermore, by analyzing the attributes such as gender, age, and occupation of the users of the stores in the same store cluster, it is possible to clarify the difference in the usage tendency of the stores. For example, it can be seen that there are many "female" and "housewives" among the users of community-based stores. On the other hand, it can be seen that there are many "males" among the users of cross-border stores.

Alternatively, by analyzing the store attributes of the store cluster, for example, a store cluster of stores handling various products such as convenience stores and a store cluster of specialty stores with limited product categories can be found. In addition, the user cluster of users who use specialty stores can be found from convenience stores and the like. By using this, for example, it is possible to expand the audience by focusing on users who use specialty stores rather than convenience stores. Therefore, targeting from each specialty store to these users is effective.

[1-3. About the application target of analysis processing]
The analysis process by the above-mentioned analysis device 10 is not limited to the classification of users or stores using the settlement history data. For example, by classifying products and users using e-commerce payment history data, the audience can be expanded for users who purchase non-standard products that reflect their hobbies and lifestyles, rather than purchasing standard products. Is possible.

Specifically, the analyzer 10 uses the degree of bias of the users of each product as a feature amount for the payment history data of a mall, an online supermarket, a mail-order site of an individual brand, a video distribution site, or the like, and a user cluster. Classify into. For example, in the analyzer 10, the horizontal axis direction of the figure shown in FIG. 2 is the product, the vertical axis direction is the user, the frequency of appearance such as the number of purchases of each product by each user, and the total number of users who have purchased each product. TFIDF is calculated as a score representing the degree of bias of users who purchase each product by using the reciprocal of (reciprocal user frequency).

Then, the analyzer 10 classifies the users showing the same feature amount into the same user cluster, using the combination of the scores of each user for each product as the feature amount for the non-standard products specified from the attribute of the product. .. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each user to the seed user, thereby extracting a user cluster consisting of users similar to the seed user.

Using this result, the analysis device 20 can find out which user clusters tend to purchase non-standard products rather than standard products. Therefore, it is possible to expand the audience as users who purchase non-standard products that reflect their hobbies and lifestyles rather than purchasing standard products, and targeting each product to these users is effective.

Alternatively, the analyzer 10 classifies the product and the user in the same manner as above by using the history data of the payment in the membership-based store such as the rental video store, so that the hobby and lifestyle are more than the purchase of the standard product. It will be possible to expand the audience for users who purchase non-standard products that reflect the above.

Specifically, the analyzer 10 classifies the history data of payments of membership-based stores such as rental video stores into user clusters by using the degree of bias of users of each product as a feature amount. For example, in the analyzer 10, the horizontal axis direction of the figure shown in FIG. 2 is the product, the vertical axis direction is the user, the frequency of appearance such as the number of purchases of each product by each user, and the total number of users who have purchased each product. TFIDF is calculated as a score representing the degree of bias of users who purchase each product by using the reciprocal of (reciprocal user frequency).

Then, the analyzer 10 classifies the users showing the same feature amount into the same user cluster, using the combination of the scores of each user for each product as the feature amount for the non-standard products specified from the attribute of the product. .. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each user to the seed user, thereby extracting a user cluster consisting of users similar to the seed user.

Using this result, the analysis device 20 can find out which user clusters tend to purchase non-standard products rather than standard products. Therefore, it is possible to expand the audience as users who purchase non-standard products that reflect their hobbies and lifestyles rather than purchasing standard products, and targeting each product to these users is effective.

Alternatively, by classifying programs and users using the history data of watching TV programs, for users who watch non-standard programs that reflect their hobbies and lifestyles rather than watching standard programs such as news. , Audience expansion is possible.

Specifically, the analyzer 10 classifies the history data of viewing a television program into a user cluster by using the degree of bias of the viewing users of each program as a feature amount. For example, in the analyzer 10, the horizontal axis direction of the figure shown in FIG. 2 is a program, the vertical axis direction is a user, the appearance frequency such as the number of times each program is viewed by each user, and the total number of users who have viewed each program. TFIDF is calculated as a score representing the degree of bias of users who watch each program by using the reciprocal of (reciprocal user frequency).

Then, the analyzer 10 classifies the users showing the same feature amount into the same user cluster by using the combination of the scores of each user for each program as the feature amount for the non-standard programs specified from the attributes of the program. .. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each user to the seed user, thereby extracting a user cluster consisting of users similar to the seed user.

Using this result, the analysis device 20 can find a user cluster that tends to watch a non-standard program rather than a standard program. Therefore, it is possible to expand the audience as users who watch non-standard programs that reflect their hobbies and lifestyles rather than watching standard programs, and targeting each program to these users is effective.

As described above, the analysis process by the analyzer 10 of the embodiment can be used or purchased for a wide variety of products, stores, contents, services, etc., and can be used or purchased by each user according to the user ID. It is applicable when historical data can be specified. In that case, for a target that is widely used or purchased by many people and a target that is used or purchased by a small number of users, the use or purchase of a small number of users should be emphasized as the feature quantity of the target. Allows users or targets to be categorized into clusters.

As a result, for example, the new service can be promoted to users who are similar to the users who used the new service, or the shopping district can be promoted to users who are similar to the users of the community-based shopping district. It is possible to effectively promote services and the like.

[2. Analysis device configuration]
Next, the above-mentioned analyzer 10 will be described with reference to FIG. FIG. 3 is a diagram showing a configuration example of the analyzer 10 according to the embodiment. As shown in FIG. 2, the analyzer 10 includes a communication unit 11, a control unit 12, and a storage unit 13.

(About communication unit 11)
The communication unit 11 is realized by, for example, a NIC (Network Interface Card) or the like, is connected to the network N by wire or wirelessly, and transmits / receives information to / from the payment server 30 and the analysis device 20.

(About storage unit 13)
The storage unit 43 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 13 of the embodiment stores the user database 13a, the payment history database 13b, and the store database 13c.

(About user database 13a)
FIG. 4 is a diagram showing an example of the user database 13a according to the embodiment. As shown in FIG. 4, the user database 13a is information in which "user ID", "attribute information", and the like are associated with each other. In addition to the information shown in FIG. 4, the user database 13a may contain arbitrary information about the user, such as posted information.

In the example shown in FIG. 4, the "user ID" is user identification information that identifies the user. "Attribute information" is information indicating various attributes such as a user's name, gender, age, hobbies and tastes, and address. For example, in the example shown in FIG. 4, the user ID "U1" and the attribute information "# U1" are associated and registered in the user database 13a. This information indicates that the attribute information of the user indicated by the user ID "U1" is "attribute information # U1".

In the example shown in FIG. 4, a conceptual value such as "attribute information # U1" is described, but in reality, a character string or a value indicating a user's attribute is registered in the user database 13a. There is.

(About payment history database 13b)
FIG. 5 is a diagram showing an example of the payment history database 13b according to the embodiment. As shown in FIG. 5, the payment history database 13b is information in which a "user ID", a "store ID", an amount of money, a "product ID", a "date and time", and the like are associated with each other. In addition to the information shown in FIG. 5, the payment history database 13b may contain arbitrary information related to the payment history, such as product features.

In the example shown in FIG. 5, the "user ID" is user identification information that identifies the user. The "store ID" is store identification information that identifies a store. The "product ID" is product identification information that identifies a product. The "amount" is the amount paid by the user to purchase the product. The "date and time" is information indicating the date and time when the user made a payment. For example, in the example shown in FIG. 5, the payment history database 13b contains the user ID "U1", the store ID "T1", the amount "100 yen", the product ID "P1", and the date and time "2020/6/30/10: "00" is associated and registered. This information indicates that the user indicated by the user ID "U1" purchased the product with the product ID "P1" at the store with the store ID "T1" at the date and time "2020/6/30/10: 00" for 100 yen. Shows.

(About store database 13c)
FIG. 6 is a diagram showing an example of the store database 13c according to the embodiment. As shown in FIG. 6, the store database 13c is information in which "store ID", "attribute information", and the like are associated with each other. In addition to the information shown in FIG. 6, arbitrary information about the store, such as mouth-to-mouth information, may be registered in the store database 13c.

In the example shown in FIG. 6, the "store ID" is the store identification information that identifies the store. "Attribute information" is information indicating various attributes such as a store name, location, and product category. For example, in the example shown in FIG. 6, the store ID “T1” and the attribute information “# T1” are associated and registered in the store database 13c. This information indicates that the attribute information of the store indicated by the store ID "T1" is "attribute information # T1".

In the example shown in FIG. 6, a conceptual value such as "attribute information # T1" is described, but in reality, a character string or a value indicating the attribute of the store is registered in the store database 13c. ing.

(About control unit 12)
Returning to the description of FIG. The control unit 12 is a controller, and for example, various programs stored in a storage device inside the analyzer 10 by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like use the RAM as a work area. It is realized by being executed. Further, the control unit 12 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. 3, the control unit 12 has an acquisition unit 12a, a classification unit 12b, and a specific unit 12c, and realizes or executes the function or operation of the analysis process described below. The control unit 12 may have other functional units.

(About acquisition unit 12a)
The acquisition unit 12a acquires the usage history of a predetermined target by the user. Here, the target is a store, a product, a service, a content, or the like. That is, the acquisition unit 12a acquires the usage history of any one of the store, the product, the service, and the program content as a predetermined target.

For example, the acquisition unit 12a acquires a history of payment processing performed by the user using the terminal device, including information indicating the store used by the user. Specifically, the acquisition unit 12a acquires payment history data from, for example, the payment server 30 or the like via the communication unit 11 and registers it in the payment history database 13b.

(About classification unit 12b)
The classification unit 12b is based on the usage history of each user, based on the appearance frequency of the history corresponding to the pair of the user and the target used by the user, and the reciprocal of the appearance frequency of the target used by all users. , Classify into similar user groups or target groups.

Specifically, the classification unit 12b calculates based on the history of stores used by each user, using the frequency of appearance of stores included in the history of users to be classified and the reciprocal of the total number of users in each store. Classify users who are similar to the criteria user based on the score.

For example, the classification unit 12b uses the appearance frequency represented by the number of times each user has used each store and the reciprocal of the total number of users who have used each store, and has a score indicating the degree of bias of the users in each store. , TFIDF is calculated.

Then, the classification unit 12b classifies the users showing the same feature amount into the same user cluster, using the combination of the scores of each user for each store as the feature amount. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each user to the seed user, thereby extracting a user cluster consisting of users similar to the seed user.

Further, the classification unit 12b calculates TFIDF based on the history of the stores used by each user, using the frequency of appearance of users included in the history of the stores to be classified and the reciprocal of the total number of stores used by each user. Based on the score such as, the stores similar to the standard store are classified.

For example, the classification unit 12b calculates TFIDF as a score indicating the degree of bias of the stores used by each user by using the frequency of appearance of each store by each user and the reciprocal of the total number of stores used by each user.

Then, the analyzer 10 classifies the stores showing the same feature amount into the same store cluster, using the combination of the scores of each user in each store as the feature amount. For example, the analyzer 10 applies logistic regression analysis to calculate the probability of similarity of each store to the seed store, thereby extracting a store cluster consisting of stores similar to the seed store.

(About the specific part 12c)
The specific unit 12c identifies a plurality of users having a similar tendency of the target to be used, or a plurality of targets having a similar tendency of the user to use, based on the history of the target used by each user. That is, the specific unit 12c specifies each user of the user group classified by the classification unit 12b or each target constituting the target group.

For example, the specific unit 12c identifies a plurality of users having similar tendencies of the stores to be used, or a plurality of stores having similar tendencies of the users to use, based on the history of the stores used by each user. That is, the specific unit 12c specifies each user of the user group classified by the classification unit 12b or each store of the store group.

Specifically, the specific unit 12c identifies a user whose tendency of the user is similar to the standard user or a store whose tendency of the user is similar to the standard store, based on the history of the store used by each user. .. That is, the specific unit 12c identifies a user group of users whose tendency of the user store is similar to the reference user, or a store group of stores whose tendency of the user user is similar to the standard store, classified by the classification unit 12b.

Alternatively, in the specific unit 12c, based on the history of the stores used by each user, the number of users common to the standard user is larger than the predetermined threshold value, or the number of users common to the standard store is predetermined. Identify stores that exceed the threshold. That is, in the specific unit 12c, the number of users in the user group in which the number of users in common with the reference store is greater than the predetermined threshold, or the number of users in common with the reference store, classified by the classification unit 12b, is greater than the predetermined threshold. Identify stores with a large number of stores.

Further, the specifying unit 12c refers to the user database 13a and extracts the attribute information of the user of each specified user cluster. Further, the specifying unit 12c refers to the store database 13c and extracts the attribute information of the store of each specified store cluster. Then, as a result of the analysis process, the specifying unit 12c outputs the users included in each classified user cluster or the stores included in each store cluster, and the attribute information of each, to, for example, the analysis device 20.

The analysis device 20 can propose sales promotion measures to each store by using the result output from the analysis device 10. For example, products and services such as promoting a new service to users who are similar to users who have used the new service, or promoting a shopping district to users who are similar to users of a community-based shopping district. It becomes possible to carry out the promotion effectively. As described above, according to the analysis process of the analyzer 10, it is possible to effectively use the history data of the use of products and services using the smartphone for the promotion of use.

[3. Processing procedure]
Next, a processing procedure executed by the analyzer 10 according to the embodiment will be described with reference to FIG. 7. FIG. 7 is a sequence diagram showing a processing procedure executed by the analyzer 10 according to the embodiment.

As shown in FIG. 7, first, in the analyzer 10, the acquisition unit 12a acquires the usage history of a predetermined target by the user. For example, the acquisition unit 12a acquires a history of payment processing performed by the user using the terminal device, including information indicating the store used by the user. Specifically, the acquisition unit 12a acquires payment history data from, for example, the payment server 30 or the like via the communication unit 11 (step S1), and registers the payment history data in the payment history database 13b.

Next, the classification unit 12b determines the appearance frequency of the history corresponding to the pair of the user and the target used by the user based on the usage history of each user, and the reciprocal of the appearance frequency of the target used by all users. Is used to classify users or target groups that are similar to each other.

Specifically, the classification unit 12b uses the appearance frequency of the stores included in the history of the users to be classified and the reciprocal of the total number of users of each store based on the payment history data of the stores used by each user. Based on the score calculated in the above, classify into user clusters that are similar to each other.

Further, the classification unit 12b calculates based on the payment history data of the stores used by each user, using the frequency of appearance of users included in the history of the stores to be classified and the reciprocal of the total number of stores used by each user. Based on the score obtained, the stores are classified into similar store clusters (step S2).

The specific unit 12c identifies a plurality of users having a similar tendency of the target to be used, or a plurality of targets having a similar tendency of the user to use, based on the history of the target used by each user. That is, the specific unit 12c specifies each user of the user group classified by the classification unit 12b or each target constituting the target group.

For example, the specific unit 12c identifies a plurality of users having similar tendencies of the stores to be used, or a plurality of stores having similar tendencies of the users to use, based on the history of the stores used by each user. That is, the specifying unit 12c identifies each user of the user cluster classified by the classification unit 12b or each store of the store cluster (step S3).

Further, the specifying unit 12c refers to the user database 13a and extracts the attribute information of the user of each specified user cluster. Further, the specifying unit 12c refers to the store database 13c and extracts the attribute information of the store of each specified store cluster. Then, as a result of the analysis process, the specifying unit 12c outputs the users included in each classified user cluster or the stores included in each store cluster and the attribute information of each to, for example, the analysis device 20 (step S4). .. This completes a series of analysis processes.

[4. effect〕
As described above, the classification device 10 according to the embodiment includes an acquisition unit 12a and a classification unit 12b. The acquisition unit 12a acquires the usage history of a predetermined target by the user. Based on the usage history by each user, the classification unit 12b uses the appearance frequency of the history corresponding to the pair of the user and the target used by the user and the reciprocal of the appearance frequency of the target used by all users. , Classify into similar user groups or target groups.

As a result, it is possible to easily classify large-scale usage history data because it is possible to classify without considering the attributes of each user and each target at the time of classification. Therefore, according to the analysis process of the analyzer 10, the history data of the use of goods, services, and the like can be effectively used for promotion of use.

Further, the specific unit 12c identifies a plurality of users having a similar tendency of the target to be used, or a plurality of targets having a similar tendency of the user to use, based on the history of the target used by each user. This makes it possible, for example, to promote the new service to users who are similar to the users who have used the new service. In addition, it will be possible to promote the shopping district to users who are similar to the users of the community-based shopping district. Alternatively, it is possible to promote products and services other than the standard products and services of e-commerce and membership services to users who have similar usage tendencies to current users. Alternatively, it is possible to promote non-standard TV programs that depend on hobbies and lifestyles to users who have similar viewing habits to current viewers. In this way, the history data of the use of products and services can be effectively used for promotion of use.

Further, the acquisition unit 12a acquires the usage history of any one of the store, the product, the service, and the program content as the target. This makes it possible to effectively promote the use of stores, products, services, program contents, and the like.

In addition, the classification unit 12b calculates a score based on the history of stores used by each user, using the frequency of appearance of stores included in the history of users to be classified and the reciprocal of the total number of users in each store. Based on, classify users who are similar to the baseline user. Further, the classification unit 12b is a score calculated by using the frequency of appearance of users included in the history of stores to be classified and the reciprocal of the total number of stores used by each user, based on the history of stores used by each user. Based on, classify stores similar to the standard store. As a result, since it is possible to classify without considering the attributes of each user and each store, it is possible to easily classify the history data of large-scale usage.

[5. Hardware configuration]
Further, the analyzer 10 according to the above-described embodiment is realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 8 is a hardware configuration diagram showing an example of a computer 1000 that realizes the functions of the analyzer 10 according to the embodiment. The computer 1000 has a CPU 1100, a RAM 1200, a ROM 1300, an HDD 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F) 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 depending 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 transmits the data generated by the CPU 1100 to the other device via the network N.

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 generated data to the output device via the input / output interface 1600.

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 analyzer 10 according to the embodiment, the CPU 1100 of the computer 1000 realizes the function of the control unit 12 by executing the program loaded on the RAM 1200. Further, the data in the storage unit 13 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.

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 modified forms.

[6. others〕
Further, among the processes described in the above-described embodiment, all or a part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed can be performed. All or part of it can be done automatically by a known method. In addition, information including processing procedures, specific names, various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the information shown in the figure.

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 them may be functionally or physically distributed / physically in any unit according to various loads and usage conditions. Can be integrated and configured.

In addition, the processes described in the above-described embodiments 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 acquisition unit 12a can be read as an acquisition means or an acquisition circuit.

10 Analytical unit 11 Communication unit 12 Control unit 12a Acquisition unit 12b Classification unit 12c Specific unit 13 Storage unit 13a User database 13b Payment history database 13c Store database 20 Analysis device

Claims (7)

An acquisition unit that acquires the usage history of a predetermined target by the user,
Based on the usage history of each user, the frequency of appearance of the history corresponding to the pair of the user and the target used by the user and the reciprocal of the appearance frequency of the target used by all users are similar to each other. A classification unit that classifies into user groups or target groups,
An analyzer characterized by having. Based on the history of the targets used by each user, a specific part that identifies multiple users with similar tendencies of the targets to be used, or multiple targets with similar tendencies of the users to be used,
The analyzer according to claim 1, further comprising. The analysis device according to claim 1 or 2, wherein the acquisition unit acquires a usage history of any one of a store, a product, a service, or a program content as the target. The classification unit is based on a score calculated using the frequency of appearance of stores included in the history of users to be classified and the reciprocal of the total number of users in each store, based on the history of stores used by each user. , The analyzer according to claim 1 or 2, characterized in that users similar to the reference user are classified. The classification unit is based on a score calculated using the frequency of appearance of users included in the history of stores to be classified and the reciprocal of the total number of stores used by each user, based on the history of stores used by each user. The analyzer according to claim 1 or 2, wherein stores similar to the reference store are classified. An analysis method performed by a computer
The acquisition process to acquire the usage history of a predetermined target by the user,
Based on the usage history of each user, the frequency of appearance of the history corresponding to the pair of the user and the target used by the user and the reciprocal of the appearance frequency of the target used by all users are similar to each other. Classification process to classify into user group or target group,
An analysis method characterized by including. Acquisition section procedure to acquire the usage history of a predetermined target by the user,
Based on the usage history of each user, the frequency of appearance of the history corresponding to the pair of the user and the target used by the user and the reciprocal of the appearance frequency of the target used by all users are similar to each other. Classification procedure to classify into user group or target group,
An analysis program characterized by having a computer execute.

Images