JP2017146683A - Generation device, generation method, and generation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate a model for appropriately predicting a predetermined object on the basis of context.SOLUTION: A generation device comprises: an acquisition part: and a generation part. The acquisition part acquires division information as action information related to an action of a user for each category related to the user, that is, action information divided on the basis of context. The generation part generates a model for predicting a predetermined object for each combination information as the division information combined on the basis of the tendency of the action information included in each division information.SELECTED DRAWING: Figure 3

Description

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

  Conventionally, a technique for predicting a predetermined target based on a generated prediction model (hereinafter simply referred to as “model”) has been provided. For example, a technique for predicting a predetermined demand using a model is provided. In such a technique, when generating a model that predicts interest in content to be distributed to users, a model is generated for all users, or a model is generated for each user.

JP 2013-196037 A

  However, in the above-described conventional technology, it may be difficult to generate a model that appropriately predicts a predetermined target. For example, when a model is generated for all of many users, it is difficult to generate a model that can be appropriately predicted for each user. For example, when a model is generated for each of a large number of users, it is difficult to generate a model that can be appropriately predicted for a user whose amount of information used for model generation is not sufficient.

  The present application has been made in view of the above, and an object thereof is to provide a generation device, a generation method, and a generation program that generate a model that appropriately predicts a predetermined target based on a context.

  The generation apparatus according to the present application includes behavior acquisition information that is behavior information related to a user's behavior for each category related to the user and that is behavior information divided based on a context, and behavior information included in each division information And a generation unit that generates a model for predicting a predetermined target for each combination information that is division information combined based on the above tendency.

  According to one aspect of the embodiment, there is an effect that it is possible to generate a model that appropriately predicts a predetermined target based on a context.

FIG. 1 is a diagram illustrating an example of a generation process according to the embodiment. FIG. 2 is a diagram illustrating an example of distribution processing according to the embodiment. FIG. 3 is a diagram illustrating a configuration example of the distribution apparatus according to the embodiment. FIG. 4 is a diagram illustrating an example of a user information storage unit according to the embodiment. FIG. 5 is a diagram illustrating an example of the behavior information storage unit according to the embodiment. FIG. 6 is a diagram illustrating an example of the division information storage unit according to the embodiment. FIG. 7 is a diagram illustrating an example of the cluster information storage unit according to the embodiment. FIG. 8 is a diagram illustrating an example of a learning information storage unit according to the embodiment. FIG. 9 is a diagram illustrating an example of a content information storage unit according to the embodiment. FIG. 10 is a flowchart illustrating an example of the generation process according to the embodiment. FIG. 11 is a flowchart illustrating an example of the generation process according to the embodiment. FIG. 12 is a flowchart illustrating an example of distribution processing according to the embodiment. FIG. 13 is a diagram illustrating an example of classifications related to users according to the embodiment. FIG. 14 is a diagram illustrating an example of the use of division information in other sections according to the embodiment. FIG. 15 is a diagram illustrating an example of use of a model in another category according to the embodiment. FIG. 16 is a diagram illustrating an example of context variation according to the embodiment. FIG. 17 is a diagram illustrating an example of context integration according to the embodiment. FIG. 18 is a diagram illustrating an example of user attribute estimation based on clusters and models according to the embodiment. FIG. 19 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the distribution apparatus.

  Hereinafter, a generation apparatus, a generation method, and a mode for executing a generation program (hereinafter referred to as “embodiment”) according to the present application will be described in detail with reference to the drawings. Note that the generation device, the generation method, and the generation program according to the present application are not limited by this embodiment. In the following embodiments, the same portions are denoted by the same reference numerals, and redundant description is omitted.

(Embodiment)
[1-1. Generation process)
First, an example of the generation process according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a generation process according to the embodiment. In the example illustrated in FIG. 1, the distribution apparatus 100 generates a model that predicts a user's interest in content as a predetermined target. The distribution apparatus 100 provides a service for distributing content to the user using the generated model, which will be described with reference to FIG. In FIG. 1 and FIG. 2, the user category is described as a user. In FIG.1 and FIG.2, the user U1-U4 shows the case where each uses terminal device 10-1 to 10-4. In addition, when demonstrating without distinguishing terminal device 10-1 to 10-4, it is named the terminal device 10 generically.

  As shown in FIGS. 1 and 2, the distribution system 1 includes a terminal device 10 and a distribution device 100. The terminal device 10 and the distribution device 100 are connected to be communicable by wire or wireless via a predetermined communication network (not shown). Note that the distribution system 1 illustrated in FIG. 1 may include a plurality of terminal devices 10 and a plurality of distribution devices 100.

  The terminal device 10 is an information processing device used by a 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. FIG. 1 shows a case where the terminal device 10 is a smartphone.

  Moreover, the terminal device 10 receives an operation by a user. In the example illustrated in FIG. 1, the terminal device 10 requests content to be displayed on a predetermined application (for example, a browser) from the distribution device 100. Hereinafter, the terminal device 10 may be referred to as a user. That is, hereinafter, the user can be read as the terminal device 10.

  The distribution apparatus 100 is a generation apparatus that generates a model that predicts a predetermined target for each combination information in which division information that is action information divided based on a context is combined. The distribution apparatus 100 is an information processing apparatus that provides a distribution service for distributing content in response to a content distribution request. For example, the distribution device 100 is an information processing device that distributes content to the terminal device 10 in accordance with a user who uses the terminal device 10 that has requested distribution of content.

  In FIG. 1, the distribution device 100 distributes content to each user and collects behavior information indicating the presence or absence of each user's behavior with respect to the distributed content. For example, the distribution apparatus 100 distributes content to the users U1 to U4, and collects behavior information indicating the presence / absence of each user's behavior with respect to the distributed content. For example, the distribution apparatus 100 collects information indicating presence / absence of a user's behavior (for example, content click) on content (for example, the content CN11) distributed to the user as behavior information.

  First, the distribution device 100 distributes content to the terminal device 10-1 used by the user U1 (step S11-1). And the delivery apparatus 100 acquires the action information which shows the presence or absence of the action of the user U1 with respect to the delivered content from the terminal device 10-1 (step S12-1). Further, the distribution device 100 distributes the content to the terminal device 10-2 used by the user U2 (step S11-2). And the delivery apparatus 100 acquires the action information which shows the presence or absence of the action of the user U2 with respect to the delivered content from the terminal device 10-2 (step S12-2). Further, the distribution device 100 distributes the content to the terminal device 10-3 used by the user U3 (step S11-3). And the delivery apparatus 100 acquires the action information which shows the presence or absence of the action of the user U3 with respect to the delivered content from the terminal device 10-3 (step S12-3). Further, the distribution device 100 distributes the content to the terminal device 10-4 used by the user U4 (step S11-4). And the delivery apparatus 100 acquires the action information which shows the presence or absence of the action of the user U4 with respect to the delivered content from the terminal device 10-4 (step S12-4).

  Note that steps S11-1 to S11-4 are for explaining the processing, and any of steps S11-1 to S11-4 may be performed first, and each of steps S11-1 to S11-4 may be performed. May be performed multiple times. Hereinafter, when it demonstrates without distinguishing step S11-1-S11-4, it is named step S11 generically. Steps S12-1 to S12-4 are for explaining the processing, and if each is after the corresponding step S11, any of steps S12-1 to S12-4 is performed first. Each step S12-1 to S12-4 may be performed a plurality of times. Hereinafter, when step S12-1 to S12-4 are described without distinction, they are collectively referred to as step S12. For example, each time step S11 is performed, step S12 may be performed to acquire action information. Further, for example, each step S11 may be performed a plurality of times, and step S12 may be performed at a predetermined timing to acquire action information.

  And the delivery apparatus 100 divides | segments the information regarding each user's action collected in step S11-S12 (step S13). Specifically, the distribution apparatus 100 divides the collected behavior information of each user based on the context.

  Here, the context means the situation of the user or the terminal device 10 and the environment (background) of the user or the terminal device 10. For example, the context is information corresponding to a situation where the terminal device 10 is used by a user or a state where a user who owns the terminal device 10 is placed. That is, the contexts in FIGS. 1 and 2 correspond to the situations of the users U1 to U4, for example, the environment where the users U1 to U4 are located. Specifically, the contexts in FIGS. 1 and 2 correspond to the situations of the users U1 to U4 when the content is distributed (displayed) to each user. For example, the context may be estimated based on the time when the content is distributed or the position information of the user acquired from the terminal device 10. Each context may be exclusive or may overlap.

  1 and 2 show contexts based on time zones and positions for the sake of simplicity. Specifically, there are four contexts in the examples of FIGS. 1 and 2: morning (4-10 o'clock), daytime (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock). Eight contexts CT11-CT18 are shown based on time zones and two locations: home (home) or office (work location).

  Here, the division information storage unit 123 in FIG. 1 shows a state in which the action information of each user is divided for each context in step S13. The division information storage unit 123 shows a case where a user as a division is a row and a context is a column. An area where one user and one context in the division information storage unit 123 intersect is defined as a square. For example, in FIG. 1, a region where the user U4 and the context CT12 intersect is defined as a mass MS42. In addition, when it demonstrates without distinguishing the position of each square, it may be set as the mass MS. In the cluster information storage unit 124 or the like, if there is an area where one user and one context intersect, the area may be a mass MS. The information included in each square MS shown in FIG. 1 is division information obtained by dividing the behavior information regarding the user's behavior for each category regarding the user based on the context.

  For example, the cell MS (upper left cell MS) at the position where the user U1 and the context CT11 intersect includes division information indicating the presence / absence of the actions AT11, AT12, AT13, and the like. For example, the cell MS at the position where the user U4 and the context CT18 intersect (the cell MS on the lower right) includes division information indicating the presence / absence of behaviors such as behaviors AT48 and AT11.

  In addition, “◯” shown on the right side of the action AT11 or the like in each cell MS in the divided information storage unit 123 indicates that the user has performed the corresponding action. In addition, “x” on the right side of the action AT11 or the like in each cell MS in the divided information storage unit 123 indicates that the user did not perform the corresponding action. For example, in the mass MS at the position where the user U1 and the context CT11 intersect, the action AT11 is “◯”, indicating that the user U1 has performed the action AT11 in the morning house. Further, for example, in the mass MS at the position where the user U1 and the context CT11 intersect, the action AT12 is “x”, indicating that the user U1 did not perform the action AT12 in the morning house. Further, for example, in the mass MS at the position where the user U1 and the context CT11 intersect, the action AT13 is “◯”, indicating that the user U1 has performed the action AT13 in the morning house.

  And the delivery apparatus 100 produces | generates a model using the division | segmentation information of each mass MS in the division | segmentation information storage part 123 (step S14). For example, the distribution apparatus 100 clusters the division information of each mass MS in the division information storage unit 123 and generates a model for each cluster. For example, the distribution apparatus 100 generates a model that predicts the user's interest in content.

  For example, the distribution apparatus 100 may perform clustering by a clustering method that does not set the number of clusters. Further, for example, the distribution apparatus 100 may perform clustering by a clustering method that sets the number of clusters. For example, the distribution apparatus 100 clusters the division information of each mass MS using a logistic regression method using a k-means method or a Dirichlet process, and generates a model using the division information for each cluster. In FIG. 1, the distribution apparatus 100 will be described below assuming that the distribution information of each mass MS is clustered using a logistic regression method using a Dirichlet process, and a model is generated using the division information for each cluster. . Details of the clustering method will be described later.

  The distribution apparatus 100 may use a mass MS including division information of a predetermined amount or more as a model generation target in step S14. For example, the distribution apparatus 100 does not target the mass MS corresponding to the context CT17 of the user U4 and the mass MS corresponding to the context CT11 of the user U2 that do not include the division information. Further, for example, the distribution apparatus 100 does not have to set the mass MS corresponding to the contexts CT13 and CT18 of the user U1 including only less than a predetermined amount (for example, two pieces) of division information as a clustering target.

  In step S14, the distribution apparatus 100 generates model information MD11. The model information MD11 includes information related to clustering of division information included in each cell MS as shown in the cluster information storage unit 124. The model information MD11 includes information on the models M1 to M6 generated for each cluster as shown in the learning information storage unit 125.

  In FIG. 1, the distribution apparatus 100 classifies the division information of each mass MS in the division information storage unit 123 into six clusters (M1 to M6 in the cluster information storage unit 124). M1 to M6 in the cluster information storage unit 124 indicate clusters to which each piece of division information belongs, and also indicate generated models. In addition, in the mass MS in the cluster information storage unit 124, a model generated by a predetermined method is assigned to the mass MS for which the codes M1 to M6 are not illustrated. For example, the distribution apparatus 100 may assign a model to a mass MS to which a model in the cluster information storage unit 124 is not assigned based on a model assigned to another mass MS. For example, the distribution apparatus 100 assigns the model M4 assigned to the mass MS corresponding to the context CT15 of the user U1 in the same twenties to the mass MS corresponding to the context CT15 of the user U4 to whom the model is not assigned. Also good. Further, for example, the distribution apparatus 100 assigns, to the mass MS corresponding to the context CT17 of the user U4 to whom the model is not assigned, the model assigned to the mass MS corresponding to the context CT17 of the same user U1 or user U2. A model based on may be assigned. In this case, the distribution apparatus 100 may assign a model obtained by combining the model M2 assigned to the mass MS corresponding to the context CT17 of the user U1 and the model M1 assigned to the mass MS corresponding to the context CT17 of the user U2. Good. For example, the distribution apparatus 100 weights each feature of the model M2 assigned to the mass MS corresponding to the context CT17 of the user U1 and each feature weight of the model M1 assigned to the mass MS corresponding to the context CT17 of the user U2. May be assigned in a predetermined ratio. For example, the distribution apparatus 100 assigns a model having a larger ratio of the model M2 assigned to the mass MS corresponding to the context CT17 of the user U1 in the twenties to the mass MS corresponding to the context CT17 of the user U4. May be. For example, the distribution apparatus 100 may assign the model M1-2 combined with the ratio of the model M1 of 0.3 and the model M2 of 0.7 to the mass MS corresponding to the context CT17 of the user U4. That is, the distribution apparatus 100 adds the weight obtained by multiplying the weight of each feature of the model M1 by 0.3 and the weight obtained by multiplying the weight of each feature of the model M2 by 0.7. 2 may be assigned to the mass MS corresponding to the context CT17 of the user U4.

  In FIG. 1, the distribution apparatus 100 corresponds to a mass MS corresponding to the context CT11 of the user U1, a mass MS corresponding to the context CT15 of the user U2, a mass MS corresponding to the context CT17 of the user U2, and a context CT13 of the user U4. The division information included in the mass MS is classified into the same cluster. In addition, the distribution apparatus 100 indicates that the model M1 has been generated using the combination information obtained by combining the division information included in the four cells MS. Further, as shown in the learning information storage unit 125, the model information regarding the model M1 has a feature 1 weight of “0.5”, a feature 2 weight of “−0.4”, and a feature 3 weight of “0. 2 "etc. The feature may be set by the administrator of the distribution apparatus 100 or may be extracted in the model generation process. For example, the feature may be a character string (keyword) such as “economy”, “smartphone”, or “soccer”. The features of each model may be the same or different.

  In addition, the distribution apparatus 100 uses the same cluster for the division information included in the mass MS corresponding to the context CT12 of the user U1, the mass MS corresponding to the context CT14 of the user U2, and the mass MS corresponding to the context CT12 of the user U3. Classify into: In addition, the distribution apparatus 100 generates the model M2 by using the combination information obtained by combining the division information included in the mass MS corresponding to the context CT14 of the user U2 and the mass MS corresponding to the context CT12 of the user U3. Indicates. Further, as shown in the learning information storage unit 125, the model information related to the model M2 has a feature 1 weight of “−0.3”, a feature 2 weight of “2.1”, and a feature 3 weight of “−0”. .2 "etc.

  As described above, in FIG. 1, the distribution apparatus 100 classifies the division information of each mass MS in the division information storage unit 123 into six clusters (union information), and uses information included in each cluster (union information). To generate models M1 to M6.

  As described above, even if the distribution apparatus 100 divides the behavior information of each section based on the context, the distribution apparatus 100 suppresses an insufficient amount of information used for generating each model by performing clustering across the sections. can do. In addition, the distribution apparatus 100 can generate an appropriate model according to each context of each section by clustering the information divided based on the section and the context. Therefore, the distribution apparatus 100 can generate a model that appropriately predicts a predetermined target based on the context.

  In addition, although the example which uses the information regarding the presence or absence of the user's action to the distributed content was shown in FIG. 1 as the action information, the action information is an action performed by the user, and what is collectable information? May be correct information. For example, the behavior information may be information related to the user's electronic commerce, or may be behavior related to a distributed advertisement (for example, click or conversion). Further, the position is not limited to a house (home) or an office (work location), but may be various information such as a transit location when commuting, a frequently visited location, or a move.

[1-2. (Delivery process)
Next, an example of distribution processing according to the embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating an example of distribution processing according to the embodiment. FIG. 2 shows a case where the user is the user U1. In FIG. 2, a case will be described as an example where the distribution apparatus 100 has already acquired user information regarding the user attributes of the user U1 and has acquired user information such as an action history.

  First, the distribution apparatus 100 acquires a content distribution request from the terminal apparatus 10-1 used by the user U1 (step S15). In FIG. 2, the terminal device 10-1 transmits information (for example, login ID etc.) for identifying the user U1, position information indicating the location of the terminal device 10-1, and the like to the distribution device 100 in step S15. The terminal device 10 acquires position information using a GPS (Global Positioning System) sensor or the like. In FIG. 2, it is assumed that step S15 is performed in a situation where the time is 9:00 am and the location (position) of the terminal device 10-1 is the position LC12-1. Note that the position LC12-1 etc., which is information indicating a position and a hyphen ("-") is attached, means position information within a predetermined range from the position LC12. Further, the positions LC11-1, LC11-2, etc. (see FIG. 5) mean position information within a predetermined range from the position LC11. That is, when the position LC12-1 indicates the position information of the user U1, the user U1 is estimated to be at the work place at the position LC12.

  The distribution apparatus 100 that has acquired the content distribution request from the terminal apparatus 10-1 determines a model for determining the contents to be distributed to the terminal apparatus 10-1 (step S16). The distribution apparatus 100 estimates that the distribution request source is the user U1 based on the information for identifying the user U1 acquired from the terminal apparatus 10-1. And the delivery apparatus 100 estimates the place where the user U1 exists based on the user information of the user U1 memorize | stored in the user information storage part 121, and position LC12-1 of the terminal device 10-1. In FIG. 2, as shown in the user information storage unit 121, the work location of the user U <b> 1 is the position LC <b> 12, and the position of the terminal device 10-1 is the position LC <b> 12-1. Estimated to be in the workplace).

  The distribution apparatus 100 estimates the context as the context CT12 corresponding to the morning office because the distribution request time in step S15 is 9:00 am. Thereafter, the distribution apparatus 100 determines the model used for determining the content to be distributed as the model M2 based on the information indicating the model M2 stored in the mass MS corresponding to the user U1 and the context CT12 in the learning information storage unit 125. To do.

  For example, the distribution apparatus 100 calculates the score of each content stored in the content information storage unit 126 (see FIG. 9) using the determined model M2. Specifically, the distribution apparatus 100 includes character information such as titles and article texts such as the contents CN11 (article A) to content CN18 (article H) stored in the content information storage unit 126, image information, and the model M2. Is used to calculate the scores of the contents CN11 to CN18 and the like. In FIG. 2, the distribution apparatus 100 calculates the scores of the contents CN11 to CN18 and the like as shown in the contents list CL11. For example, the distribution apparatus 100 calculates the score of the content CN11 that is the article A as “3.6”. For example, the distribution apparatus 100 calculates the score of the content CN12 that is the article B as “2.5”.

  And the delivery apparatus 100 determines the order | rank of each content CN11-CN18 grade | etc., Based on the score shown in the content list CL11 (step S17). In FIG. 2, the distribution apparatus 100 determines the ranks of the contents CN11 to CN18 and the like in order from the highest score shown in the content list CL11. Specifically, the distribution apparatus 100 determines the rank of the article E that is the content CN15 having the highest score as the first rank. Further, the distribution apparatus 100 determines the ranking of the article G, which is the content CN17 having the second highest score after the article E, to be the second place. Thereby, as shown in the ranking list LL11, the distribution apparatus 100 has the article E as the first rank, the article G as the second rank, the article A as the third rank, and the article F as the fourth rank. Determine that there is.

  Thereafter, the distribution apparatus 100 distributes the content shown in the ranking list LL11 to the terminal apparatus 10-1 (step S18). Specifically, the content displayed in the order of article E, article G, article A, article F, etc. is distributed to the terminal device 10-1. Then, the terminal device 10-1 that has received the content shown in the order list LL11 displays the content (step S19). In FIG. 2, the terminal device 10-1 displays an article E, an article G, an article A, and an article F that are higher-ranked contents among the contents shown in the order list LL11. The user U1 who uses the terminal device 10-1 touches the screen of the terminal device 10-1 and performs a scroll operation or the like, thereby sequentially displaying the article B, the article H, and the like below from the article F.

  As described above, the distribution apparatus 100 determines a model according to the context when the distribution request is acquired. Then, the distribution apparatus 100 determines the ranking of the content based on the content score calculated using the determined model. Thereby, the distribution apparatus 100 can distribute appropriate content according to the user and the context at the time of the distribution request. Therefore, the distribution apparatus 100 can extract appropriate content. In the above example, the case where the distribution device 100 calculates the score of each content calculated using the model selected based on the user's context, and ranks the content in order from the highest score of each calculated content. Indicated. As described above, in the above-described example, the distribution apparatus 100 uses the model that calculates the possibility of browsing each content as a continuous numerical value (score) in the context corresponding to the user. That is, in the above example, the case where the distribution apparatus 100 predicts that the higher the score of each content calculated using the model selected based on the user's context, the higher the possibility that the content will be browsed is shown. The distribution apparatus 100 may perform any prediction using the selected model.

  For example, the distribution apparatus 100 may predict that the user browses content having a score equal to or higher than a predetermined threshold and predicts that the user does not browse content having a score lower than the predetermined threshold. As described above, the distribution apparatus 100 may or may not allow the user to browse the content (for example, the value “1”) based on the score of each content calculated using the model selected based on the context (for example, the value “0”). )) May be predicted. The distribution apparatus 100 is not limited to binary prediction, and may generate a model that predicts a continuous value or a discrete value, and may predict a continuous value or a discrete value based on the generated model. For example, the distribution apparatus 100 may perform prediction corresponding to a so-called multi-label problem that handles not only the binary values “0” and “1” but also three or more values. For example, the distribution apparatus 100 may predict in which of the three or more stages the user is present, instead of the two stages of whether or not the user is viewing content. For example, the distribution apparatus 100 may predict which stage the user is likely to view the content by using a plurality of threshold values. Specifically, the prediction device 100 may use the first threshold and the second threshold smaller than the first threshold, and if the score is equal to or higher than the first threshold, the user may view the content. If the score is less than the first threshold and greater than or equal to the second threshold, it is determined that the user is likely to browse the content, and if the score is less than the second threshold, the content May be predicted to be less likely to be viewed by the user. Further, for example, the distribution apparatus 100 may generate a model that predicts the number of reactions or predicts a score of a predetermined target, and may perform prediction based on the generated model. In other words, any prediction may be performed using the model selected by the distribution apparatus 100 based on the user context. Further, the distribution apparatus 100 may generate a model according to a target to be predicted as described above.

[2. Configuration of distribution device]
Next, the configuration of the distribution apparatus 100 according to the embodiment will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the distribution apparatus 100 according to the embodiment. As illustrated in FIG. 3, the distribution apparatus 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The distribution apparatus 100 includes an input unit (for example, a keyboard and a mouse) that receives various operations from an administrator of the distribution apparatus 100 and a display unit (for example, a liquid crystal display) for displaying various types of information. May be.

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

(Storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. As illustrated in FIG. 3, the storage unit 120 according to the embodiment includes a user information storage unit 121, a behavior information storage unit 122, a division information storage unit 123, a cluster information storage unit 124, and a learning information storage unit 125. And a content information storage unit 126.

(User information storage unit 121)
The user information storage unit 121 according to the embodiment stores various types of information regarding the user. For example, the user information storage unit 121 stores various types of information regarding user attributes. FIG. 4 is a diagram illustrating an example of a user information storage unit according to the embodiment. The user information storage unit 121 illustrated in FIG. 4 includes items such as “user ID”, “age”, “sex”, “home”, “work location”, and “interest”.

  “User ID” indicates identification information for identifying a user. For example, the user identified by the user ID “U1” corresponds to the user U1 illustrated in the example of FIG. “Age” indicates the age of the user identified by the user ID. The “age” may be a specific age of the user identified by the user ID, such as 35 years old. “Gender” indicates the gender of the user identified by the user ID.

  “Home” indicates location information of the user's home identified by the user ID. In the example illustrated in FIG. 4, “home” indicates an abstract code such as “LC11”, but may be information indicating latitude and longitude. For example, “home” may be a region name or an address.

  “Work location” indicates position information of the user's work location identified by the user ID. In the example illustrated in FIG. 4, “work location” indicates an abstract code such as “LC12”, but may be information indicating latitude and longitude. For example, “home” may be a region name or an address.

  “Interest” indicates the interest of the user identified by the user ID. That is, “interest” indicates an object that is highly interested by the user identified by the user ID. In the example illustrated in FIG. 4, one “interest” is illustrated for each user, but may be plural.

  For example, in the example illustrated in FIG. 4, the age of the user identified by the user ID “U1” is “20s”, and the gender is “male”. For example, the user identified by the user ID “U1” indicates that the home is “LC11”. For example, the user identified by the user ID “U1” indicates that the work location is “LC12”. For example, the user identified by the user ID “U1” indicates that he / she is interested in “sports”.

  The user information storage unit 121 is not limited to the above, and may store various types of information according to the purpose. For example, the user information storage unit 121 may store information such as name, family structure, and income.

(Behavior information storage unit 122)
The behavior information storage unit 122 according to the embodiment stores various types of information regarding the user's behavior. FIG. 5 is a diagram illustrating an example of the behavior information storage unit according to the embodiment. For example, the behavior information storage unit 122 stores user behavior information for content distributed to the terminal device 10 of each user. The behavior information storage unit 122 illustrated in FIG. 5 includes items such as “user ID”, “behavior”, and “situation”. Further, the “behavior” includes items such as “behavior content” and “presence / absence”. The “situation” includes items such as “time” and “position”.

  “User ID” indicates identification information for identifying a user. For example, the user identified by the user ID “U1” corresponds to the user U1 illustrated in the example of FIG. The “behavior” indicates information related to the behavior of the corresponding user. The “situation” indicates information related to the situation of the corresponding user.

  “Action content” indicates identification information for identifying each action. For example, the action (action AT11) identified by the action content “AT11” corresponds to the user action (for example, click of the content) on the content (for example, the content CN11) distributed to the user. The action content is not limited to the identification information, and specific action content may be stored. “Presence / absence” indicates the presence / absence of the corresponding action content. When “presence / absence” is “1”, it indicates that the user has performed the corresponding action. Further, when “presence / absence” is “0”, it indicates that the user did not perform the corresponding action.

  “Time” indicates the time related to each action. The time “TM11” indicates the time related to the action AT11. For example, the time “TM11” may be a time when content (for example, content CN11) is distributed to the user. Further, for example, the time “TM11” may be a time when the user clicks on content (for example, content CN11) distributed to the user. The time “TM12” regarding the action (action AT12) identified by the action content “AT12” having presence / absence “0” may be, for example, the time when the content is distributed to the user. For example, the time “TM12” may be a time after a predetermined time has passed without being clicked after the content is distributed to the user.

  “Position” indicates a position related to each action. The position “LC11-1” indicates a position related to the action AT11. For example, the position “LC11-1” may be position information when the user requests distribution of content. Further, for example, the position “LC11-1” may be position information when the user clicks on the distributed content.

  For example, in the example shown in FIG. 5, the user identified by the user ID “U1” has the behavior identified by the behavior content “AT13” (the behavior AT13) is “1”, and has performed the behavior AT13. Show. Further, the user identified by the user ID “U1” indicates that the time regarding the action AT13 is the time “TM13” and the position is the position “LC11-3”.

  The behavior information storage unit 122 is not limited to the above, and may store various types of information according to the purpose. For example, the behavior information storage unit 122 may store information related to the time from when the content is distributed until the user acts. Moreover, although the case where action information was memorize | stored in the action information storage part 122 for every user ID was shown in FIG. 5, action information may be memorize | stored not only for every user ID but in order of time, for example.

(Division information storage unit 123)
The division information storage unit 123 according to the embodiment stores various types of information regarding the division information. FIG. 6 is a diagram illustrating an example of the division information storage unit according to the embodiment. The division information storage unit 123 illustrated in FIG. 6 stores action information divided based on “section” and “context”. Specifically, behavior information corresponding to each “context” is stored for each “class” of the user. For example, the information stored in the division information storage unit 123 is generated from the behavior information stored in the behavior information storage unit 122.

  “Category” indicates a division related to the user. FIG. 6 shows a case where users U1 to U4, that is, each user is a section. The “classification” is not limited to each user but may be an abstracted user classification or the like, which will be described later. “Context” indicates a context. FIG. 6 shows a context based on a time zone and a location for ease of explanation. Specifically, the “context” includes four time zones: morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), Eight contexts CT11-CT18 based on two locations: home) or office (work location) are shown. In addition, “◯” shown on the right side of the action content such as “AT11” in each square shown in FIG. 6 corresponds to “1” indicating presence / absence, and “x” corresponds to “0” indicating presence / absence. That is, in the case of “◯”, it indicates that the user has performed the corresponding action. Further, “x” indicates that the user did not perform the corresponding action.

  For example, in the example illustrated in FIG. 6, the user U1 indicates that the action AT11 is “◯” in the context CT11 and the action AT11 is performed at home (home) in the time zone “morning”. Further, the user U1 indicates that the action AT11 is “x” in the context CT12 and the action AT11 is not performed in the office (working place) in the time zone “morning”.

  In addition, the division | segmentation information storage part 123 may memorize | store not only the above but various information according to the objective. Further, in FIG. 6, for the sake of explanation, the case where information regarding the action content and presence / absence of each action information is stored in the division information storage unit 123 is illustrated, but the division information storage unit 123 is stored in the action information storage unit 122. Information regarding which cell corresponds to the action information thus performed may be stored.

(Cluster information storage unit 124)
The cluster information storage unit 124 according to the embodiment stores various types of information regarding combinations of pieces of division information. FIG. 7 is a diagram illustrating an example of the cluster information storage unit according to the embodiment. The cluster information storage unit 124 shown in FIG. 7 stores information indicating how the division information of each square shown in the division information storage unit 123 in FIG. 6 is combined and information for identifying a model corresponding to each combination. Is done. Specifically, information for identifying a model corresponding to each square based on the user's “classification” and “context” is stored. For example, the information stored in the cluster information storage unit 124 is generated by a process for generating a model.

  “Category” indicates a division related to the user. FIG. 7 shows a case where the users U1 to U4, that is, each user is a section, as in FIG. “Context” indicates a context. In FIG. 7, the context based on the time zone and the position is shown as in FIG. Specifically, the “context” includes four time zones: morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), Eight contexts CT11-CT18 based on two locations: home) or office (work location) are shown. Note that “M1”, “M2”, and the like in each square shown in FIG. 7 indicate information for identifying the model generated by the combination information obtained by combining the division information of the square in which the same code is stored.

  For example, in the example illustrated in FIG. 7, the model identified by “M1” (hereinafter, may be referred to as “model M1”. The same applies to other “M2” to “M6”) is the context CT11 of the user U1. From the combination information combining the division information included in the mass MS corresponding to the context CT15 of the user U2, the mass MS corresponding to the context CT17 of the user U2, the mass MS corresponding to the context CT17 of the user U4, and the context CT13 of the user U4 Indicates that it has been generated. Also, for example, the model M2 combines the division information included in the mass MS corresponding to the context CT12 of the user U1, the mass MS corresponding to the context CT14 of the user U2, and the mass MS corresponding to the context CT12 of the user U3. Indicates that it was generated from union information.

  The cluster information storage unit 124 is not limited to the above, and may store various types of information according to the purpose.

(Learning information storage unit 125)
The learning information storage unit 125 according to the embodiment stores information related to learning. For example, the learning information storage unit 125 stores model information generated by the generation process. FIG. 8 is a diagram illustrating an example of a learning information storage unit according to the embodiment. The learning information storage unit 125 illustrated in FIG. 8 includes items such as “feature 1” to “feature 3” corresponding to the models M1 to M6 as model information.

  For example, in the example illustrated in FIG. 8, the model information regarding the model M1 includes feature 1 weight “0.5”, feature 2 weight “−0.4”, feature 3 weight “0.2”, and the like. Indicates that For example, when the feature (feature value) of the model is expressed by an m-dimensional vector, the number of features is m, and the weights of the features 1 to m are stored.

  Note that the learning information storage unit 125 is not limited to the above, and may store various model information according to the purpose.

(Content information storage unit 126)
The content information storage unit 126 according to the embodiment stores various types of information related to content (articles). FIG. 9 is a diagram illustrating an example of a content information storage unit according to the embodiment. The content information storage unit 126 illustrated in FIG. 9 includes items such as “content ID”, “content”, and “category”.

  “Content ID” indicates identification information for identifying the content. “Content” indicates an article that is content acquired from a provider such as a so-called content provider. FIG. 9 shows an example in which conceptual information such as “article A” is stored in “content”, but in actuality, a file indicating character information, a combination of character information and an image, or a storage location thereof. Stores path names and so on. “Category” indicates classification information based on the content.

  For example, in the example illustrated in FIG. 9, the article A (content CN11) identified by the content ID “CN11” is classified into the category “economy”. Further, in the example illustrated in FIG. 9, the article B (content CN12) identified by the content ID “CN12” is classified into the category “sports”.

  The content information storage unit 126 is not limited to the above, and may store various information according to the purpose. For example, the content information storage unit 126 may store information regarding the date and time when the content was acquired and the date and time when the content was created. The content information storage unit 126 may store information regarding the evaluation value of each provider. The content information storage unit 126 may store identification information for identifying the content provider. The content information storage unit 126 may store a content score calculated by each model.

(Control unit 130)
Returning to the description of FIG. 3, the control unit 130, for example, various programs (an example of a distribution program) stored in a storage device inside the distribution apparatus 100 by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. Is realized by using the RAM as a work area. The control unit 130 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As illustrated in FIG. 3, the control unit 130 includes an acquisition unit 131, an estimation unit 132, a generation unit 133, a determination unit 134, and a distribution unit 135, and functions and functions of information processing described below. Realize or execute. Note that the internal configuration of the control unit 130 is not limited to the configuration illustrated in FIG. 3, and may be another configuration as long as the information processing described below is performed. Further, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 3, and may be another connection relationship.

(Acquisition part 131)
The acquisition unit 131 acquires user behavior information. The acquisition unit 131 acquires user behavior information from the terminal device 10 used by the user. Moreover, the acquisition part 131 acquires the user information regarding a user. In addition, the acquisition unit 131 acquires action information related to the user's action for each category related to the user and is divided information that is action information divided based on the context. For example, the acquisition unit 131 acquires division information from the division information storage unit 123. Moreover, the acquisition part 131 acquires the division | segmentation information by which the action information regarding the action of the user for every classification | category based on a user or a user attribute was divided | segmented based on the context as a classification.

  The acquisition unit 131 acquires a content distribution request from the terminal device 10. Further, the acquisition unit 131 acquires information for identifying the user (for example, login ID, terminal identification information, etc.), position information of the terminal device 10, and the like.

(Estimation unit 132)
The estimation part 132 estimates various information from the acquired various information. For example, the estimation unit 132 estimates various information based on the time when the content is distributed and the position information of the user acquired from the terminal device 10. For example, the estimation unit 132 estimates a user who is a distribution request source based on information for identifying a user acquired from the terminal device 10. For example, the estimation unit 132 estimates the location where the user is based on the user information of the user stored in the user information storage unit 121 and the position of the terminal device 10. In FIG. 2, since the work location of the user U1 is the position LC12 and the position of the terminal device 10-1 is the position LC12-1, as shown in the user information storage unit 121, the estimation unit 132 determines that the user U1 is in the office ( Estimated to be in the workplace). In FIG. 2, the estimation unit 132 estimates the context as the context CT12 corresponding to the morning office because the distribution request time is 9:00 am.

  In addition, when one division is a user, the estimation unit 132 uses the similarity between the combination information including each division information of the user and the combination information including each division information of the other division based on the similarity. The user attribute is estimated in detail later.

(Generator 133)
The generation unit 133 may generate division information. Further, the generation unit 133 may store the generated division information in the division information storage unit 123. For example, the generation unit 133 may divide the collected information regarding each user's behavior. For example, the generation unit 133 divides the collected behavior information of each user based on the context estimated by the estimation unit 132.

  The production | generation part 133 produces | generates the model which estimates a predetermined object for every combination information which is the division information combined based on the tendency of the behavior information contained in each division information. For example, in FIG. 1, the mass MS at the position where the user U1 and the context CT11 intersect and the mass MS at the position where the user U4 and the context CT13 intersect are classified into the same cluster. The cell MS at the position where the user U1 and the context CT11 intersect and the cell MS at the position where the user U4 and the context CT13 intersect each include division information whose action AT11 is “◯”. The cell MS at the position where the user U1 and the context CT11 intersect and the cell MS at the position where the user U4 and the context CT13 intersect include division information indicating the same tendency. Then, the generation unit 133 uses the combination information including the division information of the mass MS at the position where the user U1 and the context CT11 intersect and the division information of the mass MS at the position where the user U4 and the context CT13 intersect. Is generated.

  In addition, the generation unit 133 generates a model that predicts a predetermined target related to the user included in the category corresponding to the division information in the context corresponding to each of the division information included in the combination information. The production | generation part 133 produces | generates the model which predicts a predetermined | prescribed object for every union information including the combination over between divisions. For example, the generation unit 133 uses the combination information including the division information of the mass MS at the position where the user U2 and the context CT12 intersect and the division information of the mass MS at the position where the user U4 and the context CT14 intersect. Is generated. That is, the production | generation part 133 produces | generates the model M6 using the union information straddling between the divisions of the user U2 and the user U4.

  For example, the generation unit 133 generates a predetermined value for each combination information in which the division information is combined so that the prediction accuracy of a predetermined target in the combination of user classification and context corresponding to each division information included in the combination information is high. Generate a model that predicts the target. For example, the generation unit 133 generates a model for predicting a predetermined target for each combination information obtained by combining pieces of divided information with similar behavior information included in each piece of divided information. For example, the generation unit 133 generates a model that predicts the user's interest in content as the predetermined target.

  For example, the generation unit 133 generates a model for predicting a predetermined target for each combination information in which the pieces of division information are combined by a method in which the number of combination information is a predetermined number. For example, the generation unit 133 generates a model for each combination information in which the division information is combined using the k-means method. In this case, the generation unit 133 generates union information by combining the division information so that the number of clusters (number of combination information) determined (set) by the determination unit 134 is generated, and generates a model for each combination information.

  For example, the generation unit 133 generates a model for predicting a predetermined target for each combination information in which the division information is combined by a method in which the number of combination information can vary. For example, the generation unit 133 generates a model for each combination information in which the division information is combined using a logistic regression method using a Dirichlet process. For example, the production | generation part 133 produces | generates union information by combining division information sequentially, and produces | generates a model for every union information.

  For example, the production | generation part 133 produces | generates the union information which combined several division | segmentation information, and produces | generates a model with respect to the union information. For example, the generation unit 133 randomly selects a plurality of pieces of division information to generate combination information, and generates a model for the combination information. Then, the generation unit 133 verifies the prediction accuracy of the generated model. Prediction accuracy here means the probability that a user's action presence or absence can be predicted. For example, the prediction accuracy mentioned here refers to the probability that, when a model is used, for a content that is “◯” in the behavior information, a predetermined score or more is calculated. And the production | generation part 133 may perform the process which combines the division | segmentation information, maintaining the combination, when the prediction accuracy of the produced | generated model is more than a predetermined threshold value. Further, the generation unit 133 may use a predetermined hyperparameter such as a penalty term. For example, the generation unit 133 may use a penalty term that decreases in value as the number of clusters decreases. For example, the generation unit 133 may determine whether to maintain or cancel the combination based on a numerical value calculated by subtracting a value obtained by multiplying the number of clusters by a predetermined coefficient from the prediction accuracy of the model. In this case, the generation unit 133 generates combination information and models classified into a predetermined number of clusters based on a trade-off between the decrease in the number of clusters and the prediction accuracy of the model.

  For example, the generation unit 133 may verify the prediction accuracy of a model generated by using a part for generation of a model and the remaining for verification by cross validation. For example, the generation unit 133 may use 80% of the association information for model generation and the remaining 20% for verification. For example, when the verified prediction accuracy satisfies a predetermined criterion, the generation unit 133 may perform a process of maintaining the combination and further combining the division information. Moreover, the production | generation part 133 may cancel | release the combination and may perform the process which combines another division | segmentation information, when the verified prediction accuracy does not satisfy | fill a predetermined reference | standard. Further, each time the division information is combined, the generation unit 133 maintains or cancels the combination based on an average obtained by performing cross-validation (cross-validation) a plurality of times by exchanging information used for model generation and information used for verification. It may be determined.

  For example, when one division is a user and the division information of the user in one context does not satisfy a predetermined condition, the generation unit 133 divides another division similar to the user attribute of the user in the one context. A model is generated by adding information to the split information of the user in one context. For example, when the amount of user division information in one context is less than a predetermined amount, the generation unit 133 adds the division information of other sections in one context to the user division information in one context, thereby Is generated. Details of these points will be described later.

(Determining unit 134)
The determination unit 134 determines a model for determining content to be distributed to the terminal device 10. In FIG. 2, the determination unit 134 uses a model M2 as a model used for determining content to be distributed based on information indicating the model M2 stored in the mass MS corresponding to the user U1 and the context CT12 in the cluster information storage unit 124. To decide.

  For example, the determination unit 134 calculates a content score based on the determined model and content. For example, the determination unit 134 determines the rank of each content based on the calculated score. In FIG. 2, the determination unit 134 determines the ranks of the contents CN11 to CN18 and the like based on the scores shown in the content list CL11. In addition, the determination unit 134 determines the rank of the article E, which is the content CN15 having the highest score, as the first rank. Further, the determination unit 134 determines the rank of the article G, which is the content CN17 having the second highest score after the article E, to be the second place. As a result, as shown in the ranking list LL11, the determination unit 134 has the article E as the first rank, the article G as the second rank, the article A as the third rank, and the article F as the fourth rank. Determine that there is.

(Distributor 135)
The distribution unit 135 distributes various information to the terminal device 10. For example, the distribution unit 135 distributes content to the terminal device 10. The distribution unit 135 distributes the content determined by the determination unit 134. In FIG. 2, the distribution unit 135 distributes the content shown in the rank list LL11 to the terminal device 10-1. In addition, the distribution unit 135 distributes content displayed in the order of article E, article G, article A, article F, and the like to the terminal device 10-1.

[3. Generation process flow)
[3-1. (Method not to set the number of clusters)
Next, the procedure of the generation process by the distribution system 1 according to the embodiment will be described with reference to FIG. FIG. 10 is a flowchart illustrating an example of the generation process according to the embodiment. Specifically, FIG. 10 is a flowchart illustrating an example of generation processing using a technique that does not determine the number of clusters (number of combination information).

  As illustrated in FIG. 10, the acquisition unit 131 of the distribution apparatus 100 acquires the behavior information of each user (Step S101). For example, the acquisition unit 131 acquires user behavior information from the terminal device 10 used by each user.

  Further, the generation unit 133 of the distribution apparatus 100 divides information according to the context (Step S102). For example, the generation unit 133 generates division information obtained by dividing the behavior information according to the context when the behavior information is acquired. Note that step S101 and step S102 may be performed a plurality of times.

  After that, the generation unit 133 performs clustering of division information and generation of a model by a method that does not determine the number of clusters (step S103). For example, the generation unit 133 generates a model for each combination information in which the division information is combined using a logistic regression method using a Dirichlet process.

[3-2. (Method of setting the number of clusters)
Next, the procedure of the generation process by the distribution system 1 according to the embodiment will be described with reference to FIG. FIG. 11 is a flowchart illustrating an example of the generation process according to the embodiment. Specifically, FIG. 11 is a flowchart illustrating an example of a generation process using a method of setting (determining) the number of clusters (number of combination information) in advance.

  As illustrated in FIG. 11, the acquisition unit 131 of the distribution apparatus 100 acquires the behavior information of each user (Step S201). For example, the acquisition unit 131 acquires user behavior information from the terminal device 10 used by each user.

  Further, the generation unit 133 of the distribution apparatus 100 divides information according to the context (Step S202). For example, the generation unit 133 generates division information obtained by dividing the behavior information according to the context when the behavior information is acquired. Note that step S201 and step S202 may be performed a plurality of times.

  Further, the determination unit 134 of the distribution apparatus 100 sets (determines) the number of clusters (number of combination information) (step S203). For example, the determination unit 134 sets (determines) the number of clusters (number of combination information) to “6”. Note that the process of step S203 may be performed at any timing as long as it is prior to step S204.

  After that, the generation unit 133 performs clustering of division information and generation of a model by a method for determining the number of clusters (step S204). For example, the generation unit 133 generates a model for each combination information in which the division information is combined using the k-means method.

[4. (Delivery process flow)
Next, the procedure of distribution processing by the distribution system 1 according to the embodiment will be described with reference to FIG. FIG. 12 is a flowchart illustrating an example of distribution processing according to the embodiment.

  As illustrated in FIG. 12, the acquisition unit 131 of the distribution apparatus 100 acquires a distribution request (step S301). For example, the acquisition unit 131 acquires a content distribution request from the terminal device 10 used by each user.

  And the estimation part 132 of the delivery apparatus 100 estimates a user and a context (step S302). For example, the estimation unit 132 estimates a user who is a distribution request source based on information for identifying a user acquired from the terminal device 10. For example, the estimation unit 132 estimates the context based on the time when the distribution request is made and the position of the terminal device 10.

  Thereafter, the determination unit 134 of the distribution apparatus 100 determines a model (step S303). For example, the determination unit 134 determines a model for determining content to be distributed to the terminal device 10.

  Then, the determination unit 134 determines the order of content (step S304). For example, the determination unit 134 calculates a content score based on the model and content determined in step S303. For example, the determination unit 134 determines the rank of each content based on the calculated score.

  Thereafter, the distribution unit 135 of the distribution apparatus 100 distributes the content based on the order (step S305). For example, the distribution unit 135 distributes the content based on the order determined in step S304.

[5. Classification and use of information)
In the above-described example, the category related to the user has been described as a user. However, the category related to the user is not limited to the user, and various types of categories may be used depending on the purpose. This point will be described with reference to FIG. FIG. 13 is a diagram illustrating an example of classifications related to users according to the embodiment.

  Here, the division information list UL11 in FIG. 13 shows a state where the behavior information of each user is divided for each context. Similarly to the division information storage unit 123, the division information list UL11 shows a case where the user as a category is a row and the context is a column. An area where one user and one context in the division information list UL11 intersect is defined as a square. For example, in FIG. 13, an area where the user U4 and the context CT12 intersect is a mass MS42-1.

  The information included in each square MS shown in FIG. 13 is division information obtained by dividing the behavior information related to the user's behavior for each category related to the user based on the context. In FIG. 13, illustration of the action AT11 and the like is omitted. For example, in FIG. 13, “◯” and “x” in each cell MS indicate the amount of division information included in each cell MS. Further, the meanings of “◯” and “x” in each cell MS in FIG. 13 are the same as those in FIG. For example, it is indicated that the amount of division information included in the mass MS at the position where the user U1 and the context CT11 intersect is larger than the amount of division information included in the mass MS at the position where the user U1 and the context CT12 intersect. . For example, the amount of division information included in the mass MS at the position where the user U1 and the context CT12 intersect is greater than the amount of division information included in the mass MS at the position where the user U1 and the context CT13 intersect. Indicates. Further, for example, it is indicated that the division information is not included in the mass MS at the position where the user U2 and the context CT11 intersect.

  In FIG. 13, the division information for each classification based on the predetermined user attribute is generated based on the division information of each user (step S21). For example, as shown in the division information list GL11 for each classification in FIG. 13, the distribution apparatus 100 generates classification division information based on predetermined user attributes such as the classification G1, the classification G2, and the classification G3. Specifically, the distribution apparatus 100 generates division information of a male G20 classification G1. In this case, the distribution apparatus 100 integrates the division information of the user U1 and the user U4 who are males in their twenties and generates division information of the classification G1. In addition, the distribution apparatus 100 generates division information of the female group 20's classification G2. In this case, the distribution apparatus 100 integrates the division information of the user U2 who is a woman in her 20s and generates division information of the classification G2. In addition, the distribution apparatus 100 generates division information of the category G3 of the subject of interest. In this case, as shown in the user information storage unit 121, the distribution apparatus 100 generates division information of classification G3 by integrating division information such as a user U1 whose interest is a sport.

  And the delivery apparatus 100 may add the division | segmentation information of each square MS of the division | segmentation information list GL11 to a user's square MS. This point will be described with reference to FIG. FIG. 14 is a diagram illustrating an example of the use of division information in other sections according to the embodiment.

  The division information list GL11 illustrated in FIG. 14 is the same as the division information list GL11 illustrated in FIG. Here, in the division information list UL11 in FIG. 13, the cell MS42-1 at the position where the user U4 and the context CT12 intersect does not include division information. That is, the cell MS42-1 at the position where the user U4 and the context CT12 intersect is a cell MS for which the division information is insufficient.

  Therefore, the distribution apparatus 100 uses the division information for each classification (Step S22). For example, as shown in the division information list UL12, the distribution apparatus 100 adds the division information included in the corresponding mass MS in the division information list GL11 to the square MS 42-2. Specifically, the distribution apparatus 100 adds the division information included in the mass MS of the context CT12 in the classification G1 in the division information list GL11 to the mass MS 42-2. As described above, the distribution apparatus 100 adds the division information included in the corresponding mass MS in the division information list GL11 to the mass MS for which the division information is insufficient in the division information list UL11, so that the division information is included in the division information list UL11. Insufficient mass MS can be reduced. Further, for example, the distribution apparatus 100 adds the division information included in the mass MS of the context CT13 in the classification G3 corresponding to the interest “sports” of the user U1 to the mass MS42-2 of the context CT13 of the user U1. Also good. Further, the distribution apparatus 100 may add the division information included in the mass MS corresponding to the user similar to the user attribute of the user to the mass MS in which the division information of a certain user is insufficient. For example, the distribution apparatus 100 may add division information included in the mass MS of the context CT12 in the user U1 whose user attribute is similar to the mass MS42 of the context CT12 in the user U4.

  Further, the distribution apparatus 100 may generate a model using the generated division information list GL11 and use the generated model. This point will be described with reference to FIG. FIG. 15 is a diagram illustrating an example of use of a model in another category according to the embodiment. The distribution apparatus 100 may generate a model by combining the division information list UL11 and the division information list GL11. For example, the distribution apparatus 100 may perform the generation process using the division information list UL11 and the division information list GL11. For example, the distribution apparatus 100 may cluster the division information included in the division information list UL11 and the division information included in the division information list GL11 to generate a model for each cluster.

  The cluster information list ML11 in FIG. 15 includes information related to clustering of the division information included in each cell MS as shown in the cluster information storage unit 124. The cluster information list ML11 in FIG. 15 indicates that it includes six clusters indicated by M1 to M6.

  Further, the cluster information list ML12 in FIG. 15 includes information related to clustering of the division information generated from the division information list GL11 for each classification. The cluster information list ML12 in FIG. 15 indicates that it includes six clusters indicated by M11 to M13 and the like. That is, the distribution apparatus 100 generates models such as models M11 to M13 as shown in the cluster information list ML12 from the division information list GL11.

  Here, in the cluster information list ML11 in FIG. 15, the mass MS at the position where the user U4 and the context CT12 intersect does not belong to any cluster. That is, the mass MS at the position where the user U4 and the context CT12 intersect is a mass MS to which no model is assigned.

  Therefore, the distribution apparatus 100 uses a model for each classification (step S31). Specifically, the distribution apparatus 100 assigns a model included in the corresponding mass MS in the cluster information list ML12 to the mass MS at the position where the user U4 and the context CT12 intersect. In this case, the distribution apparatus 100 assigns the model M12 included in the mass MS at the position where the classification G1 and the context CT12 intersect to the mass MS at the position where the user U4 and the context CT12 intersect. Thus, the distribution apparatus 100 assigns a model in the cluster information list ML11 by assigning a model included in the corresponding mass MS in the cluster information list ML12 to a mass MS to which no model is assigned in the cluster information list ML11. Mass MS which is not done can be reduced.

[6. Context change and integration
In the example described above, the context includes four time zones: morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), and a home (home ) Or eight contexts CT11 to CT18 based on two locations in the office (work location). However, the context is not limited to the above eight but may be changed variously according to the purpose. For example, the distribution apparatus 100 may change the context according to the amount of information in each category. This point will be described with reference to FIG. FIG. 16 is a diagram illustrating an example of context variation according to the embodiment.

  The division information list UL11 illustrated in FIG. 16 is the same as the division information list UL11 illustrated in FIG. That is, the division information list UL11 includes four time zones of morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), and a house (home). Or the case where it divides | segments into eight context CT11-CT18 based on two positions of an office (work place) is shown. The division information list UL11 includes users U1 to U4 and the like.

  Here, it is assumed that the user U111 is a user with much more action information collected than the other users U1 to U4. Therefore, for the user U111, the distribution apparatus 100 generates division information based on contexts that are subdivided from the eight contexts shown in the division information list UL11.

  In FIG. 16, as shown in the division information list UL <b> 111, the distribution apparatus 100 includes 48 time zones based on 24 hours from 0:00 to 23:00 and two positions of a house (home) or an office (work location). Division information is generated based on the contexts CT111 to CT122 and the like. In other words, the distribution apparatus 100 generates the division information for the user U111 with much collected action information based on 48 contexts that are subdivided into eight contexts. Note that “0 o'clock” in the division information list UL 111 shown in FIG. 16 means the 0 o'clock range, and corresponds, for example, from 0: 0: 0 to 0:59:59. The same applies to “1 o'clock” to “23:00” in the division information list UL111 shown in FIG.

  As described above, the distribution apparatus 100 may change the context for generating the division information based on the amount of behavior information collected for each user. The distribution apparatus 100 may generate a model by combining the division information list UL11 and the division information list UL111. Moreover, for a user whose user alone includes sufficient division information in each context, such as the user U111 shown in the division information list UL111, the distribution apparatus 100 may generate a model for each of the users. For example, the distribution apparatus 100 may generate a model corresponding to each context of the user U111 using only the behavior information of the user U111 itself. Further, as illustrated in FIG. 13, the distribution apparatus 100 may add the division information of the user U111 to the user mass MS for which the division information is insufficient. That is, the distribution apparatus 100 may generate a model corresponding to each context of the user U111 and allow other users to use the information of the user U111.

  Further, when the information amount of the mass MS in the predetermined section is small, the distribution apparatus 100 may integrate the contexts in the section. This point will be described with reference to FIG. FIG. 17 is a diagram illustrating an example of context integration according to the embodiment.

  The division information list UL11 illustrated in FIG. 17 is the same as the division information list UL11 illustrated in FIG. That is, the division information list UL11 includes four time zones of morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), and a house (home). Or the case where it divides | segments into eight context CT11-CT18 based on two positions of an office (work place) is shown. The division information list UL11 includes users U1 to U4 and the like.

  Here, it is assumed that the user U112 is a user who has very little action information collected at midnight compared to the other users U1 to U4 and the like. Therefore, for the user U112, the distribution apparatus 100 generates the division information based on the seven contexts obtained by integrating the contexts related to midnight among the eight contexts shown in the division information list UL11. Specifically, the distribution apparatus 100 outputs the division information based on the seven contexts including the context CT121 corresponding to midnight, in which the context CT17 and the context CT18 related to midnight are integrated, among the eight contexts shown in the division information list UL11. Generate. That is, the distribution apparatus 100 generates the division information for the user U112 that has very little action information collected at midnight, based on the context that eliminates the distinction between home (home) and office (work location) at midnight. To do.

  As described above, the distribution apparatus 100 may integrate contexts for generating division information based on the amount of behavior information collected for each user. The distribution apparatus 100 may generate a model by combining the division information list UL11 and the division information list UL112.

[7. User attribute estimation)
In addition, the distribution apparatus 100 may estimate the user attribute of the user based on the arrangement of clusters in each category. This point will be described with reference to FIG. FIG. 18 is a diagram illustrating an example of user attribute estimation based on clusters and models according to the embodiment.

  The division information list UL21 illustrated in FIG. 18 is obtained by adding the division information of the user U21 to the division information list UL11 illustrated in FIG. That is, the division information list UL21 includes four time zones of morning (4-10 o'clock), noon (10-16 o'clock), night (16-22 o'clock), and midnight (22-4 o'clock), and a house (home). Or the case where it divides | segments into eight context CT11-CT18 based on two positions of an office (work place) is shown. Further, the division information list UL21 includes users U1 to U4, U21, and the like. Further, it is assumed that the user U21 is a user whose gender and age are user attributes.

  First, the distribution apparatus 100 generates a model using the division information of each cell MS in the division information list UL21 (step S41). For example, the distribution apparatus 100 clusters the division information of each mass MS in the division information list UL21 and generates a model for each cluster as shown in the cluster information list ML21. For example, the distribution apparatus 100 generates a model by the same process as step S14 in FIG.

  Thereafter, the distribution apparatus 100 estimates the gender and age of the user U21 whose gender and age are unknown from the cluster arrangement (step S42). For example, as shown in the user attribute information UA21, the distribution device 100 estimates the gender and age of the user whose user U21 in the division information list UL21 has a similar cluster arrangement and cluster arrangement as the user U21 gender and age. .

  Here, in FIG. 18, the distribution apparatus 100 classifies the mass MS corresponding to the context CT11 of the user U4 and the mass MS corresponding to the context CT12 of the user U21 into the same cluster. That is, the same model M3 is applied to the user U4 and the user U21 in the morning time zone. Also, the distribution apparatus 100 classifies the mass MS corresponding to the context CT13 of the user U4 and the mass MS corresponding to the context CT13 of the user U21 into the same cluster. That is, when the user U4 and the user U21 are at home (home) during the daytime, the same model M1 is applied. Further, the distribution apparatus 100 classifies the mass MS corresponding to the context CT16 of the user U4 and the mass MS corresponding to the context CT17 of the user U21 into the same cluster. That is, the same model M6 is applied to the user U4 and the user U21 in the time zone from night to midnight.

  As described above, since the user U4 and the user U21 are similar in the arrangement of clusters, that is, the arrangement of applied models, the distribution apparatus 100 sets the gender and age of the user U21 to be the same as the gender and age of the user U4. Presume that there is. Specifically, the distribution apparatus 100 estimates that the gender of the user U21 is the same “male” as the user U4, and estimates the age is the same “20s” as the user U4. As described above, the distribution apparatus 100 can estimate the user attribute of the user whose user attribute is unknown based on the cluster arrangement, that is, the similarity of the applied model arrangement.

[8. effect〕
As described above, the distribution device 100 according to the embodiment includes the acquisition unit 131 and the generation unit 133. The obtaining unit 131 obtains division information that is behavior information related to the user's behavior for each category related to the user and that is behavior information divided based on the context. Moreover, the production | generation part 133 produces | generates the model which estimates a predetermined object for every combination information which is the division information combined based on the tendency of the behavior information contained in each division information.

  Accordingly, the distribution apparatus 100 according to the embodiment can generate an appropriate model according to each context of each section by clustering information divided based on the section and context. Therefore, the distribution apparatus 100 can generate a model that appropriately predicts a predetermined target based on the context.

  Moreover, in the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model that predicts a predetermined target related to a user included in the category corresponding to the division information in the context corresponding to each of the division information included in the association information. Generate.

  Accordingly, the distribution apparatus 100 according to the embodiment can generate an appropriate model according to each context of each section by clustering information divided based on the section and context. Therefore, the distribution apparatus 100 can generate a model that appropriately predicts a predetermined target based on the context.

  Moreover, in the distribution apparatus 100 according to the embodiment, the acquisition unit 131 acquires division information obtained by dividing action information related to user behavior for each classification based on a user or a user attribute based on a context.

  Accordingly, the distribution apparatus 100 according to the embodiment can adjust the amount of division information included in each square by appropriately changing the division. Therefore, it is possible to generate a model that appropriately predicts a predetermined target based on the context.

  Moreover, in the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model for predicting a predetermined target for each combination information including a combination straddling between the sections.

  Thereby, even if the distribution apparatus 100 according to the embodiment divides the behavior information of each section based on the context, the amount of information used to generate each model is insufficient by performing clustering across the sections. Can be suppressed. Therefore, the distribution apparatus 100 can generate a model that appropriately predicts a predetermined target based on the context.

  Further, in the distribution device 100 according to the embodiment, the generation unit 133 combines the pieces of division information so that the prediction accuracy of a predetermined target in the combination of user classification and context corresponding to each piece of division information included in the combination information is high. A model for predicting a predetermined target is generated for each union information.

  Thereby, the distribution apparatus 100 according to the embodiment combines the division information so that the prediction accuracy of the predetermined target in the combination of the user classification and the context corresponding to each division information included in the combination information becomes high, thereby obtaining the context. A model that appropriately predicts a predetermined target can be generated based on the above.

  Moreover, in the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model for predicting a predetermined target for each piece of combination information in which pieces of division information similar in action information included in each piece of division information are combined.

  Thereby, the delivery apparatus 100 which concerns on embodiment can produce | generate the model which predicts a predetermined target appropriately based on a context by combining the division information with which the action information contained in each division information is similar.

  Moreover, in the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model for predicting a predetermined target for each combination information in which the division information is combined by a method in which the number of combination information is a predetermined number.

  Thereby, the delivery apparatus 100 which concerns on embodiment can produce | generate the model which predicts a predetermined target appropriately based on a context by using the method in which the number of union information becomes a predetermined number.

  Moreover, in the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model that predicts a predetermined target for each combination information in which the division information is combined by a method in which the number of combination information can be changed.

  Thereby, the delivery apparatus 100 which concerns on embodiment can produce | generate the model which predicts a predetermined target appropriately based on a context by using the method in which the number of union information can fluctuate.

  In the distribution apparatus 100 according to the embodiment, the generation unit 133 generates a model that predicts the user's interest in content as a predetermined target.

  Accordingly, the distribution apparatus 100 according to the embodiment can generate a model that appropriately predicts the user's interest in content based on the context.

  Further, in the distribution device 100 according to the embodiment, the generation unit 133, when one division is a user, when the division information of the user in one context does not satisfy a predetermined condition, the user of the user in one context A model is generated by adding the division information of another category similar to the attribute to the division information of the user in one context.

  Thereby, the distribution apparatus 100 according to the embodiment can generate a model that appropriately predicts a predetermined target based on the context, regardless of the user.

  Also, in the distribution device 100 according to the embodiment, the generation unit 133, when the amount of division information of a user in one context is less than a predetermined amount, displays the division information of another division in one context as a user in one context. A model is generated by adding to the division information.

  Accordingly, the distribution apparatus 100 according to the embodiment can generate a model that appropriately predicts a predetermined target based on the context, even if the user has insufficient division information.

  In addition, the distribution apparatus 100 according to the embodiment includes an estimation unit 132. When one division is a user, the estimation unit 132 determines the user's user based on the similarity between the combination information including each division information of the user and the combination information including each division information of the other division. Estimate attributes.

  Thereby, the distribution apparatus 100 according to the embodiment can appropriately estimate the user attribute even when the user attribute of the user is unknown.

[9. Hardware configuration)
The distribution apparatus 100 according to the embodiment described above is realized by a computer 1000 configured as shown in FIG. 19, for example. FIG. 19 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the distribution apparatus. The computer 1000 includes a CPU 1100, RAM 1200, ROM 1300, HDD 1400, communication interface (I / F) 1500, input / output interface (I / F) 1600, and media interface (I / F) 1700.

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

  The HDD 1400 stores programs executed by the CPU 1100, data used by the programs, and the like. The communication interface 1500 receives data from other devices via the network N and sends the data to the CPU 1100, and transmits data generated by the CPU 1100 to other devices via the network N.

  The CPU 1100 controls an output device such as a display and a printer and an input device such as a keyboard and a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. In addition, 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 it to the CPU 1100 via the RAM 1200. The CPU 1100 loads such a 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. Etc.

  For example, when the computer 1000 functions as the distribution device 100 according to the embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 130 by executing a program loaded on the RAM 1200. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them. However, as another example, these programs may be acquired from other devices via the network N.

  As described above, some of the embodiments and modifications of the present application have been described in detail with reference to the drawings. However, these are mere examples, and various types based on the knowledge of those skilled in the art, including the aspects described in the disclosure line of the invention. It is possible to carry out the present invention in other forms that have been modified and improved.

[10. Others]
In addition, among the processes described in the above-described embodiments and modifications, all or a part of the processes described as being automatically performed can be manually performed, or are described as being performed manually. All or part of the processing can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified. For example, the various types of information illustrated in each drawing is not limited to the illustrated information.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  In addition, the above-described embodiments and modifications can be combined as appropriate within a range that does not contradict processing contents.

  In addition, the “section (module, unit)” described above can be read as “means” or “circuit”. For example, the acquisition unit can be read as acquisition means or an acquisition circuit.

1 Distribution System 100 Distribution Device (Generation Device)
121 User Information Storage Unit 122 Behavior Information Storage Unit 123 Division Information Storage Unit 124 Cluster Information Storage Unit 125 Learning Information Storage Unit 126 Content Information Storage Unit 130 Control Unit 131 Acquisition Unit 132 Estimation Unit 133 Generation Unit 134 Determination Unit 135 Distribution Unit 10 Terminal Device N network

Claims (14)

An acquisition unit for acquiring division information that is action information related to a user's action for each category related to the user, and that is action information divided based on a context;
A generation unit that generates a model for predicting a predetermined target for each combination information that is division information combined based on a tendency of behavior information included in each division information;
A generating apparatus comprising: The generator is
The generation device according to claim 1, wherein a model for predicting a predetermined target related to a user included in a section corresponding to the division information in a context corresponding to each of the division information included in the association information is generated. . The acquisition unit
3. The generation according to claim 1, wherein the division information obtained by dividing behavior information related to a user or user behavior for each classification based on a user attribute is divided based on a context as the classification. apparatus. The generator is
The generation apparatus according to any one of claims 1 to 3, wherein a model for predicting a predetermined target is generated for each combination information including a combination straddling between the sections. The generator is
Generates a model that predicts a predetermined target for each combination information that is combined with division information so that the prediction accuracy of the predetermined target in the combination of user classification and context corresponding to each division information included in the combination information is high. The generation device according to claim 1, wherein The generator is
The model for predicting a predetermined target is generated for each combination information obtained by combining pieces of divided information with similar behavior information included in each piece of divided information. Generator. The generator is
7. The model for predicting a predetermined target is generated for each combination information in which the division information is combined by a method in which the number of the combination information is a predetermined number. 7. Generator. The generator is
The generation according to any one of claims 1 to 6, wherein a model for predicting a predetermined target is generated for each combination information in which the division information is combined by a method in which the number of combination information can vary. apparatus. The generator is
The generation apparatus according to claim 1, wherein a model that predicts a user's interest in content is generated as the predetermined target. The generator is
When one division is a user and the division information of the user in one context does not satisfy a predetermined condition, the division information of another division similar to the user attribute of the user in the one context is set to the one division. The generation apparatus according to claim 1, wherein the model is generated by adding to the division information of the user in the context of the user. The generator is
If the amount of partition information of the user in the one context is less than a predetermined amount, the model is obtained by adding the partition information of the other section in the one context to the partition information of the user in the one context. It produces | generates. The production | generation apparatus of Claim 10 characterized by the above-mentioned. When one division is a user, the user attribute of the user is estimated based on the similarity between the combination information including each division information of the user and the combination information including each division information of the other division An estimator to
The generation device according to claim 1, further comprising: A generation method executed by a computer,
An acquisition process for acquiring division information, which is action information related to a user's action for each category related to the user, and is action information divided based on a context;
A generation process for generating a model for predicting a predetermined target for each combination information that is division information combined based on a tendency of behavior information included in each division information;
A generation method comprising: An acquisition procedure for acquiring division information, which is behavior information related to the user's behavior for each category regarding the user, and is behavior information divided based on the context;
A generation procedure for generating a model for predicting a predetermined target for each combination information, which is division information combined based on the behavior information included in each division information,
A program for causing a computer to execute.

Images