JP6938259B2 - Generation device, generation method, and generation program - Google Patents

Description

The present invention relates to a generator, a generator, and a generator.

Conventionally, there has been provided a technique for predicting the presence or absence of an action related to a predetermined target based on a generated prediction model (hereinafter, simply referred to as a “model”). For example, using the information of a user who has performed an action on a predetermined target, a model for predicting whether another user will perform an action or a so-called conversion is generated.

Japanese Unexamined Patent Publication No. 2013-196037

However, it is not always possible to generate useful information with the above-mentioned prior art. For example, predicting a user's behavior or the like using a prediction model merely predicts a predetermined target such as a user's behavior, and uses information predicted for the predetermined target for another target. It is difficult to properly generate the information that enables it.

The present application has been made in view of the above, and provides a generator, a generation method, and a generation program for appropriately generating information that makes information predicted for a predetermined object available for other objects. With the goal.

The generator according to the present application includes a prediction unit that predicts information related to the probability of behavior of each user regarding a predetermined target based on the behavior information of each user, and a prediction unit that predicts the probability of each user. It is characterized by including a generation unit that generates mapping information in which each user is mapped with respect to another target different from the predetermined target based on the information related to the above.

According to one aspect of the embodiment, it is possible to appropriately generate information that makes information predicted for a predetermined target available for other targets.

FIG. 1 is a diagram showing an example of a prediction process according to an embodiment. FIG. 2 is a diagram showing an example of the generation process according to the embodiment. FIG. 3 is a diagram showing a configuration example of the generation system according to the embodiment. FIG. 4 is a diagram showing a configuration example of the generator according to the embodiment. FIG. 5 is a diagram showing an example of a learning data storage unit according to the embodiment. FIG. 6 is a diagram showing an example of a model information storage unit according to the embodiment. FIG. 7 is a diagram showing an example of the behavior information storage unit according to the embodiment. FIG. 8 is a diagram showing an example of the prediction information storage unit according to the embodiment. FIG. 9 is a diagram showing an example of a mapping target information storage unit according to the embodiment. FIG. 10 is a diagram showing an example of a mapping information storage unit according to the embodiment. FIG. 11 is a flowchart showing an example of the prediction process according to the embodiment. FIG. 12 is a flowchart showing an example of the generation process according to the embodiment. FIG. 13 is a diagram showing another mapping example. FIG. 14 is a hardware configuration diagram showing an example of a computer that realizes the function of the generator.

Hereinafter, a generator, a generation method, and a mode for carrying out the generation program (hereinafter referred to as “the 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 generator, the generation method, and the generation 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 description is omitted.

(Embodiment)
[1. Prediction processing]
First, an example of the prediction process according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram showing an example of a prediction process according to an embodiment. In the example of FIG. 1, the generation device 100 predicts a score, which is information related to the probability of an action by each user regarding a predetermined target, based on the action information of each user. In the example of FIG. 1, the generation device 100 generates prediction information indicating the probability of whether or not the user purchases the A product based on the behavior information of each user. The user behavior predicted by the generation device 100 is not limited to the above, and may be various behaviors. For example, the user behavior predicted by the generator 100 may be conversion (achievement) of various set goals including opening an account of a certain financial institution, joining a member of a certain service, and the like.

Further, the generation device 100 generates mapping information in which each user is mapped with respect to another target different from the predetermined target based on the predicted score of each user. This point will be described with reference to FIG. In the example of FIG. 2, the generator 100 maps the predicted information about the prediction target “A product purchase” with respect to the use of another target application (hereinafter, simply referred to as “application”). Generate information.

[Configuration of generation system]
First, prior to the description of FIG. 1, the generation system 1 shown in FIG. 3 will be described. As shown in FIG. 3, the generation system 1 includes a terminal device 10 and a generation device 100. The terminal device 10 and the generation device 100 are connected to each other via a predetermined network N so as to be communicable by wire or wirelessly. FIG. 3 is a diagram showing a configuration example of the generation system according to the embodiment. The generation system 1 shown in FIG. 3 may include a plurality of terminal devices 10 and a plurality of generation devices 100.

The terminal device 10 is an information processing device used by the user. The terminal device 10 is realized by, for example, a smartphone, a tablet terminal, a notebook PC (Personal Computer), a desktop PC, a mobile phone, a PDA (Personal Digital Assistant), or the like. In the example shown in 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. Specifically, FIG. 1 shows a case where the terminal device 10 is a smartphone used by a user identified by the user ID “U1” (hereinafter, may be referred to as “user U1”).

Further, in the example shown in FIG. 1, the terminal device 10 will be described as the terminal device 10-1 to 10-5 according to the user who uses the terminal device 10. For example, the terminal device 10-1 is a terminal device 10 used by the user U1. Further, for example, the terminal device 10-2 is a terminal device 10 used by the user U2. Further, in the following, when the terminal devices 10-1 to 10-5 will be described without particular distinction, they will be referred to as the terminal device 10. As described above, when "user U * (* is an arbitrary numerical value)" is described, it means that the user is a user identified by the user ID "U *". For example, when described as "user U1", that user is a user identified by the user ID "U1".

The generation device 100 is an information processing device that predicts information related to the probability of behavior of a predetermined target by each user based on the behavior information of each user. In the example of FIG. 1, the generation device 100 generates prediction information indicating the probability of whether or not the user purchases the A product based on the behavior information of each user.

In addition, the generation device 100 generates mapping information in which each user is mapped with respect to another target different from the predetermined target, based on the information related to the predicted probability of each user. In the example of FIG. 2, the generation device 100 generates mapping information in which the prediction information predicted for the prediction target “A product purchase” is mapped (mapped) with respect to another target “application (use)”.

An example of the prediction process according to the embodiment will be described with reference to FIG. FIG. 1 shows a case where the generation device 100 predicts a score indicating the probability of whether or not a user purchases the A product based on the behavior information of each user.

The generation device 100 acquires the action information of the user U1 from the terminal device 10-1 used by the user U1 (step S11-1). For example, the generation device 100 acquires the behavior information of the user U1 on the Internet (web) from the terminal device 10-1. For example, the generation device 100 acquires various behavioral information on the web, such as information indicating that the user U1 has browsed the web content and information indicating that the user U1 has performed a search using a query. For example, the generation device 100 may acquire behavior information of the user U1 in the real world (hereinafter referred to as “real space”). Further, for example, even if the generation device 100 acquires various behavioral information such as information indicating that the user U1 has purchased the B product at the A convenience store and information indicating that the user U1 has eaten or eaten at the B restaurant. good.

Further, the generation device 100 acquires action information from the terminal device 10-2 used by the user U2 (step S11-2). For example, the generation device 100 acquires the behavior information of the user U2 on the web or in the real space from the terminal device 10-2. Further, the generation device 100 acquires the action information from the terminal device 10-3 used by the user U3 (step S11-3). For example, the generation device 100 acquires the behavior information of the user U3 on the web or in the real space from the terminal device 10-3.

Further, the generation device 100 acquires the action information from the terminal device 10-4 used by the user U4 (step S11-4). For example, the generation device 100 acquires the behavior information of the user U4 on the web or in the real space from the terminal device 10-4. Further, the generation device 100 acquires the action information from the terminal device 10-5 used by the user U5 (step S11-5). For example, the generation device 100 acquires the behavior information of the user U5 on the web or in the real space from the terminal device 10-5.

Hereinafter, when steps S11-1 to S11-5 will be described without distinction, they are collectively referred to as step S11. Further, the search is not limited to steps S11-1 to S11-5, and the search for each user may be performed a plurality of times. In FIG. 1, five users U1 to U5 are illustrated, but the generation device 100 is not limited to the users U1 to U5, and the behavior information of a large number of users (for example, 1 million users, 10 million users, etc.) is obtained. get. Further, the generation device 100 may acquire the action information of each user from an external device other than the terminal device 10 used by each user. For example, the generation device 100 may acquire behavior information of each user from an external device that provides a service used by each user.

As described above, the generation device 100 collects action information of each user U1 to U5 or the like (step S12). In the example of FIG. 1, the generation device 100 stores the collected action information of each user U1 to U5 or the like in the action information storage unit 123. For example, the generation device 100 associates the action information ADT1 of the user U1 with the user U1 and stores it in the action information storage unit 123. For example, the generation device 100 associates the action information ADT2 of the user U2 with the user U2 and stores it in the action information storage unit 123.

Then, the generation device 100 predicts a score indicating the probability that each user U1 to U5 or the like purchases the A product based on the collected behavior information of the users U1 to U5 or the like. First, the generation device 100 acquires a model used for prediction (step S13). In the example of FIG. 1, the generation device 100 acquires a model corresponding to the prediction target from the model information storage unit 122. In the example of FIG. 1, the generation device 100 acquires the model M1 corresponding to the prediction target “A product purchase” from the model information storage unit 122.

As described above, when "model M * (* is an arbitrary numerical value)" is described, it indicates that the model is a model identified by the model ID "M *". For example, when described as "model M1", the model is a model identified by the model ID "M1". Further, as shown in the model information storage unit 122 in FIG. 1, the model M1 is a prediction target "A product purchase", that is, a model used for predicting the possibility of purchasing the A product, and a specific model thereof. Indicates that the data is "model data MDT1". For example, the generation device 100 inputs the user's behavior information into the model M1 and causes the model M1 to output a score indicating the probability that the user corresponding to the input behavior information purchases the A product, and the model M1 outputs the score. Based on the score to be used, the presence or absence of the user purchasing the A product is predicted. The generation of the model M1 by the generation device 100 will be described later. Further, the generation device 100 may acquire a model used for prediction from an external device.

For example, the generation device 100 inputs the behavior information of each user into the model. For example, the generation device 100 inputs the action information ADT1 to ADT5 and the like of each user U1 to U5 and the like into the model M1.

In the example of FIG. 1, the generation device 100 calculates a score indicating the probability of purchasing the A product of the user U1 by the processing as shown in the processing group PS14-1. The generation device 100 inputs the action information ADT1 corresponding to the user U1 into the model M1 (step S14-1). For example, the generation device 100 inputs the behavior information of the user U1 on the web or in the real space into the model M1. The model M1 to which the action information ADT1 is input outputs a score (step S15-1). In the example of FIG. 1, the model M1 to which the behavior information ADT1 is input outputs a score “0.8” as shown in the score SC11-1.

Further, the generation device 100 calculates a score indicating the probability of the user U2 purchasing the A product by the processing as shown in the processing group PS14-2. The generation device 100 inputs the action information ADT2 corresponding to the user U2 into the model M1 (step S14-2). For example, the generation device 100 inputs the behavior information of the user U2 on the web or in the real space into the model M1. The model M1 to which the action information ADT2 is input outputs a score (step S15-2). In the example of FIG. 1, the model M1 in which the behavior information ADT2 is input outputs a score “0.55” as shown in the score SC11-2.

Further, the generation device 100 also processes the action information ADT3 to ADT5 and the like of the users U3 to U5 and the like in the same manner. For example, the generation device 100 calculates the score corresponding to each user U3 to U5 by inputting each of the action information ADT3 to ADT5 into the model M1. In the example of FIG. 1, the generation device 100 calculates the score "0.75" for the user U3, calculates the score "0.80" for the user U4, and the score "0. 04 ”is calculated.

Then, the generation device 100 predicts whether or not to purchase the A product of each user U1 to U5 or the like based on the score of each user U1 to U5 or the like (step S16). For example, the generation device 100 uses a predetermined threshold value and predicts the presence or absence of an action of each user based on the comparison between the predetermined threshold value and the score. In the example of FIG. 1, when the score of the user is equal to or higher than the threshold value "0.7", the generation device 100 predicts that the user (user with behavior) purchases the A product. Further, when the score of the user is less than the threshold value "0.7", the generation device 100 predicts that the user (user with no action) does not purchase the A product. In the example of FIG. 1, for simplification of explanation, an example of classifying users into behavioral users and behavioral non-users by comparison with a threshold value is shown, but the generation device 100 determines the probability of each user. The following processing may be performed using the score which is related information.

As a result, the generation device 100 generates prediction information indicating whether or not each user has purchased the A product, as shown in the prediction information storage unit 124 in FIG. For example, the generation device 100 predicts that the user U1 will purchase the A product because the score of the user U1 is "0.8" and the threshold value is "0.7" or more. Further, for example, the generation device 100 predicts that the user U2 will not purchase the product A because the score of the user U2 is "0.55" and is less than the threshold value "0.7". The threshold value is not limited to "0.7" and may be various values such as "0.65" and "0.8".

As described above, the generation device 100 predicts information related to the probability of behavior regarding a predetermined target of the user based on the behavior information of each user. In the example of FIG. 1, the generation device 100 inputs the behavior information of each user into the model M1, and causes the model M1 to output a score indicating the probability that each user purchases the A product. Then, when the score output by the model M1 is equal to or higher than a predetermined threshold value, the generation device 100 predicts that the user will purchase the A product. In this way, the generation device 100 can predict whether or not to perform an action related to a predetermined target for a user who has not actually performed an action related to the predetermined target by using the action information of each user and the model M1. can. Therefore, the generation device 100 amplifies the users associated with the presence / absence of the action by predicting the action even if the number of users who can acquire the information indicating the presence / absence of the action such as a predetermined conversion is small. Can be done.

[2. Generation process]
Next, an example of the generation process according to the embodiment will be described with reference to FIG. FIG. 2 is a diagram showing an example of the generation process according to the embodiment. In the example of FIG. 2, the generation device 100 generates mapping information in which the prediction information predicted for the prediction target “A product purchase” is mapped (mapped) with respect to another target “application (use)”.

First, the generation device 100 acquires the behavior prediction information (step S21). In the example of FIG. 2, the generation device 100 acquires information indicating whether or not each user U1 to U5 or the like is related to the purchase of the A product from the prediction information storage unit 124.

Further, the generation device 100 acquires the mapping target information (step S22). In the example of FIG. 2, information about the mapping target such as each user U1 to U5 is acquired from the mapping target information storage unit 125.

The “mapping target” shown in the mapping target information storage unit 125 in FIG. 2 indicates a target to which the prediction information is mapped. The “user ID” shown in the mapping target information storage unit 125 in FIG. 2 indicates identification information for identifying a user. The “type” shown in the mapping target information storage unit 125 in FIG. 2 indicates the type in the mapping target of the corresponding user behavior. The “date and time” shown in the mapping target information storage unit 125 in FIG. 2 indicates the date and time when the corresponding user action was performed.

For example, the user U1 indicates that the application of the type "A application (game)" was used in the period of the date and time dt11-dt12. That is, it indicates that the user U1 used the A application, which is a game application, between 1 o'clock and 3 o'clock on Monday. Further, for example, the user U2 indicates that he / she used the application of the type "C application (route)" in the period of the date / time dt21-dt22. That is, it indicates that the user U2 used the C application, which is a route application, between 0:30 and 2:00 on Monday. The generation device 100 may acquire user behavior information regarding the application to be mapped in step S11.

Then, the generation device 100 generates mapping information (step S23). In the example of FIG. 2, the generation device 100 uses the information shown in the prediction information storage unit 124 and the mapping target information storage unit 125 to apply the prediction information predicted for the prediction target “A product purchase” to another target “app”. Generate mapping information that is mapped (mapped) with respect to "use)". Specifically, the generation device 100 generates information in which the distribution of users is mapped for each division divided into time and day of the week, as shown in the mapping information MP11 in FIG.

Each division divided by the row "time" and the column "day of the week" in the mapping information MP11 in FIG. 2 indicates the division to be mapped. “Time” indicates the time corresponding to the segment to be mapped. In the example of FIG. 2, the "time" is "0-4", "4-8", "8-12", "12-16", "16-20", and "20-24". It is shown that it is divided into 6 pieces at the time. "Day of the week" indicates the day of the week corresponding to the category to be mapped. In the example of FIG. 2, "day of the week" is divided into seven "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", and "Sunday". That is, in the example of FIG. 2, the generation device 100 maps information for each of 42 (= 6 * 7) divisions, which is a combination of six time zones and seven days of the week.

In the example of Fig. 2, the division where the row of time "0-4" and the column of day of the week "Monday" intersect (square "Monday 0-4") is the application between 0:00 and 4:00 on Monday. Indicates the content rate, which is the ratio of users (users with behavior) who are predicted to purchase the A product among the users who used the product. In the example of FIG. 2, it is shown that the content rate of the behavioral user in the user who used the application from 0:00 to 4:00 on Monday is "5%".

For example, the generation device 100 calculates the content rate as described above by using the number of users stored in each category as shown in the mapping information storage unit 126 (see FIG. 10). For example, the generator 100 is the number of users who are predicted to purchase the A product stored in the square "Monday 0-4" and who used the application between 0:00 and 4:00 on Monday. Using (the number of users with behavior) and the total number of users who used the app between 0:00 and 4:00 on Monday, the behavior of the users who used the app between 0:00 and 4:00 on Monday. Calculate the content rate of users.

Specifically, in the square "Monday 0-4" in FIG. 10, the number of active users is "500" and the total number of users is "10000". Therefore, the generator 100 is set on Monday. The content rate of behavioral users among users who used the application between 0:00 and 4:00 is calculated as "5 (= 500/10000 * 100)%".

In the example of Fig. 2, the division where the row of time "0-4" and the column of day of the week "Tuesday" intersect (square "Tuesday 0-4") is the application between 0:00 and 4:00 on Tuesday. Indicates the content rate, which is the ratio of users (users with behavior) who are predicted to purchase the A product among the users who used the product. In the example of FIG. 2, it is shown that the content rate of the behavioral user in the user who used the application from 0:00 to 4:00 on Tuesday is "3%".

Further, in the example of FIG. 2, the division (square "Saturday 20-24") where the row of the time "20-24" and the column of the day of the week "Saturday" intersect is between 20:00 and 24:00 on Saturday. Indicates the content rate, which is the ratio of users (users with behavior) who are predicted to purchase the A product among the users who used the application. In the example of FIG. 2, it is shown that the content rate of the behavioral user in the user who used the application from 20:00 to 24:00 on Saturday is "12%".

Further, the generation device 100 shows in a display mode such as a heat map how many users who are likely to purchase the A product in each category are included according to the content rate of each category. In the example of FIG. 2, the darker the hatching, the higher the content rate of the user who purchases the A product. In the example of FIG. 2, the hatching of the category having the content rate of 8% or more is the darkest. Further, in the example of FIG. 2, the hatching of the category in which the content rate is 4% or more and less than 8% is the second darkest. Further, in the example of FIG. 2, the hatching of the category in which the content rate is 2% or more and less than 4% is the third darkest. Further, in the example of FIG. 2, the category in which the content rate is less than 2% is not hatched.

In this way, the generation device 100 generates the mapping information MP11 that shows each category in a heat map according to the content rate of the users with behavior, so that the ratio of users who purchase the A product is high in which category. Can be intuitively grasped. For example, in the example of FIG. 2, since the classification corresponding to the day of the week "Sunday" has the highest hatching, the content rate of the user who purchases the A product is given to the user who uses the application on Sunday by the mapping information MP11. It becomes possible to intuitively grasp that the value is high. For example, the generation device 100 may provide the generated mapping information MP11 to a predetermined business operator such as a store that sells the A product.

As described above, the generator 100 maps each user with respect to another target application (use) different from the A product (purchase) based on the predicted score indicating the purchase probability of the A product of each user. Generate mapping information. In the example of FIG. 2, the generation device 100 aggregates the users who used the application for each category, which is a combination for each day of the week and time zone, and uses the ratio of users with behavior among all users included in each category as the content rate. Generate mapping information mapped to each category. In this way, the generation device 100 generates mapping information indicating which category contains a large number of behavioral users based on the score indicating the predicted behavior probability, thereby providing information predicted to a predetermined target. Can be made available for the subject. Therefore, the generation device 100 can appropriately generate information that makes information predicted for a predetermined object available for other objects.

(2-1. Mapping information)
Further, in the example of FIG. 2, an example of mapping the content rate for each category based on the combination of the day of the week and the time zone is shown, but the mapping information is not limited to the content rate and may be various information. For example, the generation device 100 may generate mapping information in which expected values, average probabilities, medians, modes, and the like are mapped for each category. For example, the generation device 100 uses a score indicating the probability of all users included in a certain category to calculate a score indicating the average probability of the category, a median value, a mode value, and the like, and generates mapped mapping information. May be good.

(2-2. Category)
(2-2-1. Other information)
Further, in the example of FIG. 2, an example of mapping for each division based on the combination of the day of the week and the time zone is shown, but the division may be various information not limited to the day of the week and the time zone. For example, the generation device 100 may perform mapping for each division related to the user's position information. For example, the generation device 100 may perform mapping for each division corresponding to an area divided for each predetermined area. For example, the generation device 100 may perform mapping for each of the areas AR21 to AR24 and the like, as in the mapping information MP21 shown in FIG. FIG. 13 is a diagram showing another mapping example. Specifically, FIG. 13 shows mapping information in which the mapping category is changed from "day of the week x time zone" to "area (position)" when the purchase of product A is mapped to the application.

In the example of FIG. 13, the generation device 100 shows how many users who are likely to purchase the A product in each area are included according to the content rate of each area (classification), such as a heat map. Indicated by. In the example of FIG. 13, the darker the hatching, the higher the content rate of the user who purchases the A product. In the example of FIG. 13, as shown in the mapping information MP21, it is shown that the content rate is high in the area near the boundary line including the area AR22 and the area AR23.

In this way, the generation device 100 generates the mapping information MP21 that shows each category in the form of a heat map according to the content rate of the users with behavior, so that the user who purchases the A product in which area (category) You can intuitively grasp whether the ratio is high. For example, in the example of FIG. 13, since the area (division) located near the boundary line has the darkest hatching, the mapping information MP21 is used to purchase the A product for the user who is using the application near the boundary line. It becomes possible to intuitively grasp that the content rate of the user is high. For example, the generation device 100 may provide the generated mapping information MP21 to a predetermined business operator such as a store that sells the A product.

Further, the generation device 100 may generate mapping information using any combination of divisions. For example, the generation device 100 may generate mapping information that is mapped for each division that combines time and position (area). For example, the generation device 100 may generate mapping information mapped for each division that combines a day of the week, a time zone, and an area.

(2-2-2. Subdivision)
Further, the generation device 100 may generate mapping information for each division in which the mapping target is subdivided. For example, the generation device 100 may generate mapping information for each category in which the application to be mapped is subdivided into categories. For example, the generation device 100 may generate mapping information for each category of the application to be mapped, which is subdivided into categories such as "game", "weather", and "route". Further, for example, the generation device 100 may generate mapping information for each division of the application to be mapped for each specific application name (content). For example, even if the generation device 100 generates mapping information for the application to be mapped, the mapping information is subdivided for each specific application name such as "A application", "B application", and "C application". good.

(2-3. Other objects)
In the example of FIG. 2, the case where the mapping information is generated by targeting the application and dividing the combination of the day of the week and the time zone is shown. For example, the generation device 100 targets the day of the week and the time zone and generates the mapping information. It may be generated. In this case, for example, the generation device 100 may generate mapping information for a combination of a day of the week and a time zone. For example, the generation device 100 may generate mapping information regarding behavioral users included in all users included in each combination of day of the week and time zone.

Further, in the example of FIG. 2, the case of generating the mapping information mapped for the application is shown, but for example, the generation device 100 may generate the mapping information mapped for various targets. For example, the generation device 100 may generate mapping information that maps various targets such as a user's position information, a user's attribute information, a user's interest, a query used by the user, and a site browsed by the user.

(2-4. Behavior information)
For example, when the generation device 100 generates the model M1 used in the example of FIG. 1, the generation device 100 may generate the model M1 in which the behavior information excluding the behavior information corresponding to the correct answer information is input. For example, the generation device 100 may generate the model M1 by removing the action information corresponding to the correct answer information “purchase of product A” from the action information. Further, for example, when the generation device 100 specifies the feature amount (feature) of the model M1, the feature amount of the model M1 is set so as not to include the feature amount corresponding to the "purchase of product A" which is the correct answer information. You may. As a result, the generation device 100 can accurately predict when the user's behavior information that does not include the behavior information regarding the purchase of the A product is input to the model M1. The details of model generation will be described later.

For example, when the generation device 100 generates the model M1 used in the example of FIG. 1, the generation device 100 may generate the model M1 in which the behavior information excluding the behavior information corresponding to the mapping target is input. For example, the generation device 100 may generate the model M1 by removing the behavior information corresponding to the application to be mapped from the behavior information. Further, for example, when the generation device 100 specifies the feature amount (feature) of the model M1, the feature amount of the model M1 may be set so as not to include the feature amount corresponding to the application to be mapped. As a result, the generation device 100 can accurately predict when the user's behavior information that does not include the behavior information about the application to be mapped is input to the model M1. Further, in this case, the generation device 100 may remove the behavior information related to the application from the behavior information of the user and input it to the model M1.

[3. Generator configuration]
Next, the configuration of the generation device 100 according to the embodiment will be described with reference to FIG. FIG. 4 is a diagram showing a configuration example of the generator according to the embodiment. As shown in FIG. 4, the generation device 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The generation device 100 has an input unit (for example, a keyboard, a mouse, etc.) that receives various operations from the administrator of the generation device 100, and a display unit (for example, a liquid crystal display, etc.) for displaying various information. You may.

(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 N by wire or wirelessly, and transmits / receives information to / from the terminal device 10.

(Memory 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. As shown in FIG. 4, the storage unit 120 according to the embodiment includes a learning data storage unit 121, a model information storage unit 122, an action information storage unit 123, a prediction information storage unit 124, and a mapping target information storage unit 125. And a mapping information storage unit 126.

(Learning data storage unit 121)
The learning data storage unit 121 according to the embodiment stores various information related to the learning data. FIG. 5 is a diagram showing an example of a learning data storage unit according to the embodiment. For example, the learning data storage unit 121 stores teacher data used for generating a model. The learning data storage unit 121 shown in FIG. 5 includes items such as "data ID", "user ID", "correct answer information", and "behavior information".

The "data ID" indicates identification information for identifying the data. For example, the "data ID" indicates identification information for identifying a combination of user behavior information and correct answer information used as learning data (teacher data). The "user ID" indicates identification information for identifying a user. "Correct answer information" indicates the correct answer information corresponding to the data identified by the data ID. For example, the "correct answer information" indicates the presence or absence of the actual action of the user corresponding to the data identified by the data ID. In FIG. 5, a character string including “yes” or “no” such as “yes (1)” or “no (0)” is illustrated in the “correct answer information”. For example, the user performs an action related to a predetermined target. If so, the flag corresponding to the presence of action (for example, "1") is stored. Further, for example, when the user does not perform an action related to a predetermined target, a flag corresponding to no action (for example, "0") is stored in the "correct answer information". In FIG. 5, the “correct answer information” indicates a case where the correct answer information indicating whether or not the product A has been purchased is stored.

The "behavior information" indicates the user's behavior information corresponding to the data identified by the data ID. In FIG. 5, an example in which conceptual information such as "ADT101" is stored in "behavior information" is shown, but in reality, various information about the behavior performed by the user or a file path name indicating the storage location thereof. Etc. are stored.

For example, in the example shown in FIG. 5, the data (learning data DT1) identified by the data ID "DT1" is information (data) corresponding to the user (user U101) identified by the user ID "U101". Is shown. Further, the learning data DT1 indicates that the correct answer information is "Yes (1)". That is, the learning data DT1 indicates that the user U101 has purchased the A product. Further, it is shown that the learning data DT1 includes the behavior information ADT101.

The learning data storage unit 121 is not limited to the above, and may store various information depending on the purpose. For example, the learning data storage unit 121 may store information regarding the date and time when the learning data was added. For example, when the correct answer information is "Yes (1)", the learning data storage unit 121 may store information indicating the date and time when the user performed the action. Further, for example, the learning data storage unit 121 may store information indicating by what kind of determination processing each learning data is added. For example, the learning data storage unit 121 may store information indicating whether or not each learning data is determined by the selection of the administrator.

(Model information storage unit 122)
The model information storage unit 122 according to the embodiment stores information about the model. For example, the model information storage unit 122 stores model information (model data) generated by the generation process. FIG. 6 is a diagram showing an example of a model information storage unit according to the embodiment. The model information storage unit 122 shown in FIG. 6 includes items such as “model ID”, “prediction target”, and “model data”. Although only the models M1, M2, and M3 are shown in FIG. 6, a large amount of model information such as M4 and M5 may be stored according to each prediction target.

The "model ID" indicates identification information for identifying the model. For example, the model identified by the model ID "M1" corresponds to the model M1 shown in the example of FIG. “Prediction target” indicates the prediction target of the corresponding model. Further, "model data" indicates data of the associated corresponding model. For example, "model data" includes information including nodes in each layer, functions adopted by each node, connection relationships of nodes, and connection coefficients set for connections between nodes.

For example, in the example shown in FIG. 6, in the model (model M1) identified by the model ID "M1", the prediction target is "A product purchase", and the user corresponding to the input behavior information purchases the A product. It is shown that it is used to predict the information related to the probability of doing. Further, it is shown that the model data of the model M1 is the model data MDT1.

The model M1 (model data MDT1) includes an input layer into which user behavior information is input, an output layer, and a first element that is any layer from the input layer to the output layer and belongs to a layer other than the output layer. , A second element whose value is calculated based on the first element and the weight of the first element, and each element belonging to each layer other than the output layer with respect to the area information input to the input layer is first. By performing an operation based on the first element and the weight of the first element as an element, the value of the score used for predicting the information related to the probability of the action of the user regarding a predetermined target is output from the output layer. It is a model for making a computer work.

Here, it is assumed that the models M1 to M3 and the like are realized by the regression model represented _{by "y = a 1} * x ₁ + a ₂ * x ₂ + ... + a _i * x _i". In this case, for example, the first element included in the model M1 corresponds to input data (xi) such as x1 and x2. Further, the weight of the first element corresponds to the coefficient ai corresponding to xi. Here, the regression model can be regarded as a simple perceptron having an input layer and an output layer. When each model is regarded as a simple perceptron, the first element corresponds to any node of the input layer, and the second element can be regarded as the node of the output layer.

Further, it is assumed that the models M1 to M3 and the like are realized by a neural network having one or a plurality of intermediate layers such as DNN. In this case, for example, the first element included in the model M1 corresponds to either the node of the input layer or the intermediate layer. Further, the second element corresponds to a node in the next stage, which is a node to which a value is transmitted from a node corresponding to the first element. Further, the weight of the first element corresponds to a connection coefficient which is a weight considered for the value transmitted from the node corresponding to the first element to the node corresponding to the second element.

The model information storage unit 122 is not limited to the above, and may store various model information depending on the purpose.

(Behavior information storage unit 123)
The behavior information storage unit 123 according to the embodiment stores various information related to the user's behavior. FIG. 7 is a diagram showing an example of the behavior information storage unit according to the embodiment. For example, the behavior information storage unit 123 stores various behavior information such as the behavior of each user. The action information storage unit 123 shown in FIG. 7 includes items such as "user ID" and "behavior information".

The "user ID" indicates identification information for identifying a user. For example, the user identified by the user ID "U1" corresponds to the user U1 shown in the example of FIG. The "behavior information" indicates the user's behavior information corresponding to the data identified by the data ID. In FIG. 7, an example in which conceptual information such as "ADT1" is stored in "behavior information" is shown, but in reality, various information about the behavior performed by the user or a file path name indicating the storage location thereof. Etc. are stored.

For example, in the example shown in FIG. 7, it is shown that the action information ADT1 has been collected for the user (user U1) identified by the user ID "U1". For example, the behavior information ADT1 may include the behavior information of the user U1 on the Internet (web). For example, the behavior information ADT1 includes information indicating that the user U1 browsed the contents of the C site at a certain date and time (date and time X), and information indicating that the search was performed using the query "A product" at a certain date and time. Various web behavioral information may be included. Further, for example, the behavior information ADT1 includes various information such as information indicating that the user U1 purchased the B product at the A convenience store on a certain date and time, and information indicating that the user U1 ate and drank at the B restaurant on a certain date and time. Information may be included.

The behavior information storage unit 123 is not limited to the above, and may store various information depending on the purpose. Further, although FIG. 7 shows a case where the action information is stored in the action information storage unit 123 for each user ID, the action information is not limited to each user ID and may be stored, for example, in chronological order.

For example, the action information may include information on items such as "action ID", "date and time", "type", and "content". In this case, for example, the "behavior ID" indicates information that identifies the user's behavior. For example, "date and time" indicates the date and time when the corresponding user's action was performed. For example, a specific date and time such as "July 15, 2017 18:31:52" may be stored in the "date and time". Further, "type" indicates information regarding the type of action of the corresponding user. In addition, "content" indicates the content targeted in the corresponding user's behavior. For example, for user U1, "date and time X" is stored in the date and time in the action log identified by the action ID "AC11", "browsing" is stored in the type, and "contents of C site" is stored in the content. May be done.

(Prediction information storage unit 124)
The prediction information storage unit 124 according to the embodiment stores various information related to prediction. FIG. 8 is a diagram showing an example of the prediction information storage unit according to the embodiment. The prediction information storage unit 124 shown in FIG. 8 has items such as “prediction target”, “user ID”, “score”, and “presence / absence”. The prediction information storage unit 124 shown in FIG. 8 shows information regarding the possibility of performing the behavior of the prediction target such as the users U1 to U5 predicted in FIG.

“Prediction target” indicates a prediction target that predicts the behavior of the user. The "user ID" indicates identification information for identifying a user. For example, the user identified by the user ID "U2" corresponds to the user U2 shown in the example of FIG. The "score" indicates a score corresponding to the probability of performing the prediction target of the corresponding user. “Presence / absence” indicates a prediction of the presence / absence of user behavior. For example, “presence / absence” indicates a prediction of the presence / absence of user behavior based on a comparison between a predetermined threshold value and a score. In FIG. 8, “presence / absence” is stored as “yes (1)” when the score is equal to or higher than the threshold value “0.7”, and “none (0)” when the score is less than the threshold value “0.7”. ) ”Is stored.

In FIG. 8, a character string including "yes" or "no" such as "yes (1)" or "no (0)" is shown in "presence or absence". For example, the user performs an action related to a predetermined target. If it is predicted, a flag corresponding to the presence of action (for example, "1") is stored. Further, for example, when it is predicted that the user does not perform an action related to a predetermined target, a flag corresponding to no action (for example, "0") is stored in "presence / absence". In FIG. 8, “presence / absence” indicates a case where information indicating the prediction of purchase / non-purchase of product A is stored.

For example, in the example shown in FIG. 8, it is shown that the target for predicting the presence or absence of action is the purchase of product A. The score indicating the action probability of the user U1 is "0.8". Further, since the score "0.8" of the user U1 is equal to or higher than the threshold value "0.7", it is predicted that the presence or absence of the user U1 is "Yes (1)", that is, the purchase of the A product is performed. Is shown.

The prediction information storage unit 124 is not limited to the above, and may store various information depending on the purpose. The prediction information storage unit 124 may store a specific name or the like of the product.

(Mapping target information storage unit 125)
The mapping target information storage unit 125 according to the embodiment stores various information related to the user's behavior. FIG. 9 is a diagram showing an example of a mapping target information storage unit according to the embodiment. For example, the mapping target information storage unit 125 stores various behavioral information such as the behavior of each user. The mapping target information storage unit 125 shown in FIG. 9 includes items such as “mapping target”, “user ID”, “type”, and “date and time”.

“Mapping target” indicates a target to which the prediction information is mapped. The "user ID" indicates identification information for identifying a user. For example, the user identified by the user ID "U3" corresponds to the user U3 shown in the example of FIG. Further, the "type" indicates the type in the mapping target of the corresponding user behavior. The "date and time" indicates the date and time when the corresponding user's action was performed. In the example of FIG. 9, the period of "dt11-dt12 (Monday 1:00 to 3:00)" is shown, but the "date and time" is "July 15, 2017 19:31:52" or the like. Time point may be remembered.

For example, in the example shown in FIG. 9, it is shown that the mapping target is an “application”. The user (user U1) identified by the user ID "U1" indicates that he / she used the application of the type "A application (game)" in the period of the date / time dt11-dt12. That is, it indicates that the user U1 used the A application, which is a game application, between 1 o'clock and 3 o'clock on Monday.

The mapping target information storage unit 125 is not limited to the above, and various information may be stored depending on the purpose. Further, although FIG. 9 shows a case where the action information is stored in the mapping target information storage unit 125 for each user ID, the action information is not limited to each user ID and may be stored, for example, in chronological order.

(Mapping information storage unit 126)
The mapping information storage unit 126 according to the embodiment stores various information related to mapping. FIG. 10 is a diagram showing an example of a mapping information storage unit according to the embodiment. For example, the mapping information storage unit 126 stores mapping information for each division divided by day of the week and time.

"Mapping target" indicates a mapping target. For example, in the example shown in FIG. 10, it is shown that the mapping target is an “application”. Each section divided by the row "time" and the column "day of the week" indicates the section to be mapped.

“Time” indicates the time corresponding to the segment to be mapped. In the example of FIG. 10, the "time" is "0-4", "4-8", "8-12", "12-16", "16-20", and "20-24". It is shown that it is divided into 6 pieces at the time. The time division is not limited to the above, and may be divided into various types, for example, 24 times such as "0" time, "1" time, and "2" time.

"Day of the week" indicates the day of the week corresponding to the category to be mapped. In the example of FIG. 10, "day of the week" is divided into seven "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", and "Sunday". The day of the week is not limited to the above, and may be divided into various types, for example, "holiday (Saturday, Sunday)" and "weekday (Monday to Friday)".

For example, in the example of FIG. 10, the division (square "Monday 0-4") where the row of time "0-4" and the column of day of the week "Monday" intersect is between 0:00 and 4:00 on Monday. Information indicating the number of users who used the application is stored in. In the example of FIG. 10, in the square "Monday 0-4", the number of users who are predicted to purchase product A and who used the application between 0:00 and 4:00 on Monday (behavior). Number of users) is stored. Further, in the example of FIG. 10, the total number of users who used the application between 0:00 and 4:00 on Monday is stored in the square "Monday 0-4".

Specifically, in the square "Monday 0-4" in FIG. 10, information indicating that the number of active users is "500" and the total number of users is "10000" is stored. .. That is, information indicating that the content rate of behavioral users among users who used the application between 0:00 and 4:00 on Monday is "5 (= 500/10000 * 100)%" is stored. ..

The mapping information storage unit 126 is not limited to the above, and may store various information depending on the purpose.

(Control unit 130)
Returning to the description of FIG. 4, the control unit 130 is a controller, and is stored in a storage device inside the generation device 100 by, for example, a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). It is realized by executing various programs (corresponding to an example of a prediction program and a generation program) using the 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). The control unit 130 is any one of an input layer, an output layer, and an input layer to an output layer into which user action information is input by information processing according to a model M1 or the like stored in the model information storage unit 122. The action input to the input layer, including the first element which is a layer of the above and belongs to a layer other than the output layer, and the second element whose value is calculated based on the weights of the first element and the first element. Information related to the probability of an action by a user regarding a predetermined target by performing an operation based on the weights of the first element and the first element, with each element belonging to each layer other than the output layer as the first element. The score value used for the prediction of is output from the output layer.

As shown in FIG. 4, the control unit 130 includes an acquisition unit 131, a prediction unit 132, a generation unit 133, and a provision unit 134, and realizes or executes an information processing function or operation described below. .. 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. The control unit 130 includes a second element whose value is calculated based on the first element and the weight of the first element by information processing according to the model M1 (model data MDT1) stored in the storage unit 120. By performing an operation based on the first element and the weight of the first element with each element belonging to each layer other than the output layer as the first element for the action information input to the input layer including The value of the score used for predicting the information related to the probability of behavior regarding the target of is output from the output layer.

(Acquisition unit 131)
The acquisition unit 131 acquires various types of information. For example, the acquisition unit 131 is derived from the learning data storage unit 121, the model information storage unit 122, the action information storage unit 123, the prediction information storage unit 124, the mapping target information storage unit 125, the mapping information storage unit 126, and the like. Acquire various information. Further, the acquisition unit 131 may acquire various information from an external information processing device. Further, the acquisition unit 131 may acquire various information from the terminal device 10 or the like. For example, the acquisition unit 131 acquires the user's action information from the terminal device 10.

In the example of FIG. 1, the acquisition unit 131 acquires the action information of the user U1 from the terminal device 10-1 used by the user U1. For example, the acquisition unit 131 acquires the behavior information of the user U1 on the Internet (web) from the terminal device 10-1. For example, the acquisition unit 131 acquires various behavioral information on the web, such as information indicating that the user U1 has browsed the web content and information indicating that the user U1 has performed a search using a query. For example, the acquisition unit 131 may acquire the behavior information of the user U1 in the real space. For example, the acquisition unit 131 may acquire various behavioral information such as information indicating that the user U1 has purchased the B product at the A convenience store and information indicating that the user U1 has eaten or drink at the B restaurant.

For example, the acquisition unit 131 acquires action information from the terminal device 10-2 used by the user U2. For example, the acquisition unit 131 acquires the behavior information of the user U2 on the web or in the real space from the terminal device 10-2. For example, the acquisition unit 131 acquires action information from the terminal device 10-3 used by the user U3. For example, the acquisition unit 131 acquires the behavior information of the user U3 on the web or in the real space from the terminal device 10-3.

For example, the acquisition unit 131 acquires action information from the terminal device 10-4 used by the user U4. For example, the acquisition unit 131 acquires the behavior information of the user U4 on the web or in the real space from the terminal device 10-4. For example, the acquisition unit 131 acquires action information from the terminal device 10-5 used by the user U5. For example, the acquisition unit 131 acquires the behavior information of the user U5 on the web or in the real space from the terminal device 10-5.

In the example of FIG. 1, the acquisition unit 131 acquires a model used for prediction. For example, the acquisition unit 131 acquires a model corresponding to the prediction target from the model information storage unit 122. For example, the acquisition unit 131 acquires the model M1 corresponding to the prediction target “A product purchase” from the model information storage unit 122.

In the example of FIG. 2, the acquisition unit 131 acquires the behavior prediction information. For example, the acquisition unit 131 acquires information indicating whether or not each user U1 to U5 or the like is related to the purchase of the A product from the prediction information storage unit 124. For example, the acquisition unit 131 acquires the mapping target information. For example, the acquisition unit 131 acquires information about the mapping target such as each user U1 to U5 from the mapping target information storage unit 125.

(Prediction unit 132)
The prediction unit 132 predicts various types of information. The prediction unit 132 predicts various information using the model stored in the model information storage unit 122. For example, the prediction unit 132 predicts various information based on various information acquired by the acquisition unit 131. By inputting the user's behavior information into the model, the prediction unit 132 predicts the information related to the probability of the user's behavior regarding a predetermined target.

The prediction unit 132 predicts information related to the probability of behavior of a predetermined target by each user based on the behavior information of each user. For example, the prediction unit 132 predicts information related to the probability of an action related to a predetermined target by each user based on the behavior information of each user who has performed an action related to another target. For example, the prediction unit 132 predicts information related to the probability of an action by each user regarding a predetermined target based on the behavior information of each user other than the behavior related to another target.

In the example of FIG. 1, the prediction unit 132 predicts a score indicating the probability that each user U1 to U5 or the like purchases the A product based on the collected behavior information of the users U1 to U5 or the like. For example, by inputting the user's behavior information into the model M1, the prediction unit 132 causes the model M1 to output a score indicating the probability that the user corresponding to the input behavior information purchases the A product, and the model M1 outputs the score. Based on the score to be used, the presence or absence of the user purchasing the A product is predicted.

In the example of FIG. 1, the prediction unit 132 predicts whether or not to purchase the A product of each user U1 to U5 or the like based on the score of each user U1 to U5 or the like. For example, the prediction unit 132 predicts the presence or absence of an action of each user based on the comparison between the predetermined threshold value and the score using a predetermined threshold value. For example, the prediction unit 132 predicts that the user will purchase the A product when the score of the user is equal to or higher than the threshold value “0.7”. For example, if the user's score is less than the threshold value "0.7", the prediction unit 132 predicts that the user will not purchase the A product.

For example, the prediction unit 132 predicts that the user U1 will purchase the A product because the score of the user U1 is "0.8" and the threshold value is "0.7" or more. For example, the prediction unit 132 predicts that the user U2 will not purchase the A product because the score of the user U2 is "0.55" and is less than the threshold value "0.7".

For example, the prediction unit 132 calculates the score using a model having an arbitrary structure such as the regression model and the neural network described above. Specifically, the model M1 is a value that quantifies the prediction of the user's behavior regarding a predetermined target when the user's behavior information (that is, each element used for calculating the score described above) is input (that is, the model M1). , A coefficient is set to output a score that suggests the probability of action by the user on a given target. The prediction unit 132 uses such a model M1 to calculate a score regarding the probability of an action by the user regarding a predetermined target.

In the above example, the model M1 is a model that outputs a quantified value of the prediction of the probability of the behavior of a predetermined target by the user when the behavior information of the user is input. However, the model (model X) according to the embodiment may be a model generated based on the result obtained by repeating the input / output of data to the model M1. For example, the model X may be a model (model Y) trained to input the user's behavior information and output the score output by the model M1. Alternatively, the model M1 may be a model trained to input the user's behavior information and output the output value of the model Y. Further, when the prediction unit 132 performs prediction processing using GAN (Generative Adversarial Networks), the model M1 may be a model forming a part of GAN.

(Generator 133)
The generation unit 133 generates various information. Further, the generation unit 133 sets each user to another target different from the predetermined target based on the prediction information stored in the prediction information storage unit 124 and the mapping target information stored in the mapping target information storage unit 125. Generate the mapped mapping information.

In addition, the generation unit 133 generates mapping information in which each user is mapped with respect to another target different from the predetermined target, based on the information related to the probability of each user predicted by the prediction unit 132. For example, the generation unit 133 generates mapping information including information on the number of target users whose probability-related information is a user who satisfies a predetermined condition among each user. For example, the generation unit 133 generates mapping information including information on the number of target users whose probability-related information is equal to or greater than a predetermined threshold among each user. For example, the generation unit 133 generates mapping information including an index value based on the number of target users in each user. For example, the generation unit 133 generates mapping information including a content rate indicating the ratio of the target user to each user.

For example, the generation unit 133 generates mapping information for each division divided based on a predetermined standard. For example, the generation unit 133 generates mapping information for each division divided based on time. For example, the generation unit 133 generates mapping information for each division divided based on the day of the week. For example, the generation unit 133 generates mapping information for each division divided based on the position.

In the example of FIG. 1, the generation device 100 generates prediction information indicating whether or not each user has purchased the A product, as shown in the prediction information storage unit 124.

In the example of FIG. 2, the generation unit 133 generates mapping information. For example, the generation unit 133 uses the information shown in the prediction information storage unit 124 and the mapping target information storage unit 125 to map the prediction information predicted for the prediction target “A product purchase” with respect to another target “app” (mapping). ) Generate mapping information. For example, the generation unit 133 generates information that maps the distribution of users for each division divided into time and day of the week, as shown in the mapping information MP11 in FIG.

Further, for example, the generation unit 133 uses the learning data stored in the learning data storage unit 121 to generate a model as shown in the model information storage unit 122. For example, the generation unit 133 generates a model used for predicting information related to the probability of an action regarding a predetermined target by each user based on the learning data acquired by the acquisition unit 131. For example, the generation unit 133 determines the probability of an action by each user regarding a predetermined target based on learning data including behavior information and correct answer information indicating the presence or absence of an action regarding a predetermined target by each user corresponding to the behavior information. Generate a model used to predict relevant information.

For example, the generation unit 133 generates models M1 to M3 and the like, and stores the generated models M1 to M3 and the like in the model information storage unit 122. The generation unit 133 may generate models M1 to M3 and the like using any learning algorithm. For example, the generation unit 133 generates models M1 to M3 and the like by using learning algorithms such as a neural network, a support vector machine (SVM), clustering, and reinforcement learning. As an example, when the generation unit 133 generates models M1 to M3 and the like using a neural network, the models M1 to M3 and the like include an input layer containing one or more neurons and an intermediate layer containing one or more neurons. It has an output layer containing the above neurons.

The generation unit 133 generates a model and stores the generated model in the model information storage unit 122. Specifically, the generation unit 133 is a first element that is any of an input layer into which user action information is input, an output layer, and a layer from the input layer to the output layer, and belongs to a layer other than the output layer. And the second element whose value is calculated based on the first element and the weight of the first element, and each element belonging to each layer other than the output layer with respect to the action information input to the input layer is the first. A model that outputs a score value used for predicting information related to the probability of an action by a user regarding a predetermined target from an output layer by performing an operation based on the first element and the weight of the first element as one element. To generate.

For example, the generation unit 133 generates a model based on the training data. For example, the generation unit 133 generates a model based on the training data. For example, the generation unit 133 generates a model by performing learning using the data DT1 to DT4 in the learning data storage unit 121 as learning data (teacher data).

For example, the generation unit 133 generates the model M1 by using the data DT1 to DT4 in the learning data storage unit 121. For example, the generation unit 133 outputs the model M1 when the behavior information ADT101 included in the data DT1 is input to the model M1 when the correct answer information is "1" indicating that the product A has been purchased. The learning process is performed so that the score approaches "1". Further, for example, when the correct answer information is "0" indicating that the product A has not been purchased, the generation unit 133 causes the model M1 to input the action information ADT 102 included in the data DT2 into the model M1. The learning process is performed so that the output score approaches "0".

For example, the generation unit 133 generates a model using the feature amount related to the user's behavior information. Further, for example, the generation unit 133 generates a model by using the feature amount related to the behavior information of the user on the Web. For example, the generation unit 133 generates a model using the feature amount related to the behavior information of the user in the real space. In this way, the generation unit 133 learns the features included in the user's behavior information and generates a model that predicts the information related to the probability of the behavior regarding the user's predetermined target. The features included in the action information learned by the generation unit 133 may be input to the generation unit 133 by a person such as an administrator of the generation unit 133, or may be automatically learned (extracted) by the generation unit 133. May be good.

The model learning method is not limited to the above-mentioned method, and any known technique can be applied. It should be noted that the generation of each model may be performed by appropriately using various conventional techniques related to machine learning. For example, the model may be generated by using a technique related to machine learning of supervised learning such as SVM (Support Vector Machine). Further, for example, model generation may be performed using a technique related to machine learning of unsupervised learning. For example, model generation may be performed using a technique of deep learning. For example, the model may be generated by appropriately using various deep learning techniques such as DNN (Deep Neural Network), RNN (Recurrent Neural Network), and CNN (Convolutional Neural Network). The description regarding the generation of the model is an example, and the generation of the model may be performed by a learning method appropriately selected according to the information that can be acquired and the like. That is, if the generation unit 133 can learn the model M1 so as to output the score corresponding to the correct answer information when the behavior information included in the learning data is input, the generation unit 133 generates the model M1 by any method. May be done.

(Providing section 134)
The providing unit 134 provides various information. For example, the providing unit 134 provides various information to the terminal device 10. The providing unit 134 may provide a service based on the prediction information predicted by the prediction unit 132. For example, the providing unit 134 may provide information about the model generated by the generating unit 133 to an external information processing device. Further, for example, the providing unit 134 may provide the information output by the model to an external information processing device. For example, the providing unit 134 provides the mapping information generated by the generating unit 133 to an external information processing device.

[4. Prediction processing flow]
Next, the procedure of the generation process by the generation system 1 according to the embodiment will be described with reference to FIG. FIG. 11 is a flowchart showing an example of the prediction process according to the embodiment.

As shown in FIG. 11, the generation device 100 acquires the behavior information of each user (step S101). In the example of FIG. 1, the generation device 100 acquires the action information of each user U1 to U5 or the like from the terminal devices 10-1 to 10-5 or the like. For example, the generation device 100 acquires the behavior information of each user U1 to U5 as shown in the behavior information storage unit 123.

Further, the generation device 100 acquires a model corresponding to the prediction target (step S102). In the example of FIG. 1, the generation device 100 acquires the model M1 corresponding to the prediction target “A product purchase” from the model information storage unit 122.

Further, the generation device 100 predicts the presence / absence of an action regarding the prediction target of each user based on the action information and the model (step S103). In the example of FIG. 1, in the generation device 100, each user U1 to U5 purchases the A product as shown in the prediction information storage unit 124 by using the behavior information of each user U1 to U5 and the model M1. Predict if.

[5. Generation process flow]
Next, the procedure of the generation process by the generation system 1 according to the embodiment will be described with reference to FIG. FIG. 12 is a flowchart showing an example of the generation process according to the embodiment.

As shown in FIG. 12, the generation device 100 acquires the behavior prediction information (step S201). In the example of FIG. 2, the generation device 100 acquires the prediction information indicating whether or not to purchase the A product from the prediction information storage unit 124.

After that, the generation device 100 acquires the mapping target information (step S202). In the example of FIG. 2, the generation device 100 acquires information about the application to be mapped from the mapping target information storage unit 125.

After that, the generation device 100 generates mapping information in which the behavior prediction information is mapped with respect to the mapping target (step S203). In the example of FIG. 2, the generation device 100 generates mapping information MP11 indicating the content rate of the user who is predicted to purchase the A product among the application users in the time zone of each day of the week.

[6. effect〕
As described above, the generation device 100 according to the embodiment has a prediction unit 132 and a generation unit 133. The prediction unit 132 predicts information related to the probability of behavior of a predetermined target by each user based on the behavior information of each user. In addition, the generation unit 133 generates mapping information in which each user is mapped with respect to another target different from the predetermined target, based on the information related to the probability of each user predicted by the prediction unit 132.

As described above, the generation device 100 according to the embodiment generates mapping information in which each user is mapped with respect to another target different from the predetermined target based on the information related to the predicted probability of each user. Information that makes the information expected for a given target available for other targets can be appropriately generated.

Further, in the generation device 100 according to the embodiment, the prediction unit 132 predicts information related to the probability of the action of each user regarding a predetermined target based on the action information of each user who has performed an action related to another target. ..

As described above, the generation device 100 according to the embodiment predicts the information related to the probability of the action regarding the predetermined target by each user based on the action information of each user who has performed the action regarding the other target. Information that makes the information expected for a given target available for other targets can be appropriately generated.

Further, in the generation device 100 according to the embodiment, the prediction unit 132 predicts information related to the probability of an action related to a predetermined target by each user based on the action information of each user other than the action related to another target.

As described above, the generation device 100 according to the embodiment is determined by predicting the information related to the probability of the action regarding the predetermined target by each user based on the action information of each user other than the action regarding the other target. It is possible to appropriately generate information that makes the information predicted for one object available for other objects.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information including information on the number of target users whose probability-related information satisfies a predetermined condition among the users.

As described above, the generation device 100 according to the embodiment is determined by generating mapping information including information on the number of target users whose probability-related information is a user satisfying a predetermined condition among each user. Information that makes the information expected of a target available to other targets can be appropriately generated.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information including information on the number of target users whose probability-related information is equal to or higher than a predetermined threshold value among each user.

As described above, the generation device 100 according to the embodiment generates mapping information including information on the number of target users whose probability-related information is equal to or higher than a predetermined threshold value, thereby generating information predicted for a predetermined target. Information that can be made available for other objects can be appropriately generated.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information including an index value based on the number of target users in each user.

As described above, the generation device 100 according to the embodiment can use the information predicted for a predetermined target for other targets by generating mapping information including an index value based on the number of target users in each user. Information can be generated appropriately.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information including the content rate indicating the ratio of the target users in each user.

As described above, the generation device 100 according to the embodiment can use the information predicted for a predetermined target for other targets by generating the mapping information including the content rate indicating the ratio of the target users in each user. Information can be generated appropriately.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information for each division divided based on a predetermined standard.

As described above, the generation device 100 according to the embodiment makes the information predicted for the predetermined target available to the other target by generating the mapping information for each of the divisions divided based on the predetermined standard. Information can be generated appropriately.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information for each division divided based on time.

In this way, the generation device 100 according to the embodiment generates mapping information for each division divided based on time, so that information predicted for a predetermined target can be used for other targets. Can be generated appropriately.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information for each division divided based on the day of the week.

In this way, the generation device 100 according to the embodiment generates mapping information for each division divided based on the day of the week, so that information predicted for a predetermined target can be used for other targets. Can be generated appropriately.

Further, in the generation device 100 according to the embodiment, the generation unit 133 generates mapping information for each division divided based on the position.

In this way, the generation device 100 according to the embodiment generates mapping information for each division divided based on the position, so that information predicted for a predetermined target can be used for other targets. Can be generated appropriately.

[7. Hardware configuration]
The generation device 100 according to the above-described embodiment is realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 14 is a hardware configuration diagram showing an example of a computer that realizes the function of the generator. 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 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 an 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 generator 100 according to the embodiment, the CPU 1100 of the computer 1000 is controlled by executing a program or data (for example, model M1 (model data MDT1)) loaded on the RAM 1200. The function of the unit 130 is realized. The CPU 1100 of the computer 1000 reads and executes these programs or data (for example, model M1 (model data MDT1)) from the recording medium 1800, but as another example, these programs from another device via the network N. May be obtained.

Although some of the embodiments and modifications of the present application have been described in detail with reference to the drawings, these are examples and vary based on the knowledge of those skilled in the art, including the embodiments described in the disclosure line of the invention. It is possible to carry out the present invention in another form in which the above is modified or improved.

[8. 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 are 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 dispersed / physically distributed in any unit according to various loads and usage conditions. Can be integrated and configured.

Further, 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 acquisition unit can be read as an acquisition means or an acquisition circuit.

1 Generation system 100 Generation device 121 Learning data storage unit 122 Model information storage unit 123 Behavior information storage unit 124 Prediction information storage unit 125 Mapping target information storage unit 126 Mapping information storage unit 130 Control unit 131 Acquisition unit 132 Prediction unit 133 Generation unit 134 Provider 10 Terminal device N network

Claims (13)

Based on the behavior information of each user, a prediction unit that predicts a score, which is information related to the probability of behavior of each user regarding a predetermined target, and a prediction unit.
Mapping information that maps each user with respect to another target different from the predetermined target based on the score of each user predicted by the prediction unit, and is the mapping information of the user who has performed an action regarding the other target. Among them, a generation unit that generates mapping information indicating the ratio of users whose score is equal to or higher than a predetermined threshold value.
A generator characterized by comprising. The prediction unit
The generator according to claim 1 , wherein the score of the behavior of each user regarding a predetermined target is predicted based on the behavior information of each user who has performed an action related to the other target. The prediction unit
It said other of said non-action on the object based on the action information of each user, the generating device according to claim 1, characterized in that predicting the score of action on a given subject by each user. The generator
The generation according to any one of claims 1 to 3, wherein the mapping information including information on the number of target users whose scores satisfy a predetermined condition among the users is generated. Device. The generator
The generation device according to claim 4, wherein the mapping information including information regarding the number of target users whose score is equal to or higher than a predetermined threshold value among the users is generated. The generator
The generator according to claim 4 or 5, wherein the mapping information including an index value based on the number of target users in each user is generated. The generator
The generator according to any one of claims 4 to 6, wherein the mapping information including the content rate indicating the ratio of the target user in each user is generated. The generator
The generator according to any one of claims 1 to 7, wherein the mapping information is generated for each division divided based on a predetermined standard. The generator
The generation device according to claim 8, wherein the mapping information is generated for each division divided based on time. The generator
The generator according to claim 8 or 9, wherein the mapping information is generated for each division divided based on the day of the week. The generator
The generator according to any one of claims 8 to 10, wherein the mapping information is generated for each division divided based on the position. It ’s a computer-executed generation method.
A prediction process for predicting a score, which is information related to the probability of behavior of a predetermined target by each user based on the behavior information of each user.
Mapping information that maps each user with respect to another target different from the predetermined target based on the score of each user predicted by the prediction step, and is the mapping information of the user who has performed an action regarding the other target. Among them, a generation step of generating mapping information indicating the ratio of users whose score is equal to or higher than a predetermined threshold value, and
A generation method characterized by including. Based on the behavior information of each user, a prediction procedure for predicting a score, which is information related to the probability of behavior of each user regarding a predetermined target, and a prediction procedure.
Mapping information that maps each user with respect to another target different from the predetermined target based on the score of each user predicted by the prediction procedure, and is the mapping information of the user who has performed an action regarding the other target. A generation procedure for generating mapping information indicating the percentage of users whose score is equal to or higher than a predetermined threshold value.
A generator characterized by having a computer execute.

Images