JP7715607B2 - Update device, update method, and update program - Google Patents

Description

The present invention relates to an update device, an update method, and an update program.

Conventionally, for example, a learning device has been proposed that learns a model that generates vectors representing the characteristics of a user from the user's online behavior history (see, for example, Patent Document 1).

Japanese Patent Application Laid-Open No. 2019-125313

However, conventional technology leaves room for improvement, for example, when it comes to updating vectors. Specifically, vectors are generated or updated based on online behavior, and no consideration has been given to updating vectors based on behavior in the real world.

The present invention has been made in consideration of the above, and aims to provide an update device, update method, and update program that can appropriately update vectors that indicate characteristics.

To solve the above-mentioned problems and achieve the objectives, the update device of the present invention comprises a linking unit that links a user to the areas visited by the user based on the user's visit history in a bipartite graph including a set of users and a set of areas, and an update unit that updates the user vector or the area vector based on a user vector that indicates the characteristics of the user based on the user's behavior on the network and an area vector that indicates the characteristics of the areas linked to the user by the linking unit.

According to the present invention, the vector representing the features can be appropriately updated.

FIG. 1 is a diagram illustrating an example of information processing according to an embodiment. FIG. 2 is a block diagram illustrating an example of the configuration of an update device according to the embodiment. FIG. 3 is a diagram illustrating an example of a user vector storage unit according to the embodiment. FIG. 4 is a diagram illustrating an example of a region vector storage unit according to the embodiment. FIG. 5 is a schematic diagram of a user vector and a region vector according to the embodiment. FIG. 6 is a schematic diagram illustrating an example of updating a region vector according to the embodiment. FIG. 7 is a flowchart illustrating an example of an update process according to the embodiment. FIG. 8 is a hardware configuration diagram illustrating an example of a computer that realizes the functions of the update device according to the embodiment.

Below, the update device, update method, and update program according to the present application will be described in detail with reference to the drawings. Note that the update device, update method, and update program according to the present application are not limited to these embodiments.

[Embodiment]
[1. Information Processing]
First, an example of information processing according to the embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram illustrating an example of information processing according to the embodiment.

As shown in FIG. 1, the update device 10 according to the embodiment acquires network behavior information and real-world behavior information of a user (not shown) and outputs a user vector or a region vector.

Here, online behavioral information refers to, for example, information about the usage history of various portal sites, including purchase history, search history, browsing history, etc. Real-world behavioral information refers to, for example, information about each user's real-world visit history, including, for example, information about the location information acquired by the user's user device (e.g., a smartphone).

A user vector is a vector that indicates the characteristics of a user based on that user's behavior on the network. For example, a user vector is generated by a user vector conversion model that is trained to generate user vectors that are more similar the more similar the behavioral information on the network. Note that vectors being similar refer to vectors whose cosine distances are similar.

Area vectors are vectors that indicate the characteristics of real-world areas. For example, area vectors are generated by an area vector conversion model that has been trained to output vectors that are more similar the closer the areas themselves are, the similarity of the facilities located there, the similarity of the roads, and the similarity of the visiting users' attributes. Areas are, for example, divided areas obtained by dividing map information according to a predetermined rule (for example, a mesh). Therefore, in the bipartite graph described below, areas are also connected according to their geographical location.

First, the update device 10 according to the embodiment acquires network behavior information and real-world behavior information for each user (steps S01 and S02). Next, the update device 10 connects nodes in a bipartite graph including a set of users and a set of regions (step S03).

Here, a bipartite graph is a graph in which each user and each region serves as a node. For example, a node corresponding to a user stores the user vectors Vu1-Vu3 of the corresponding user, and a node corresponding to a region stores the region vectors Va1-Va3 of the corresponding region.

The update device 10 connects, with edges, the node corresponding to a user and the node corresponding to the area visited by the user based on the user's visit history (real-world behavioral information). Furthermore, in the bipartite graph, areas are connected according to their geographical locational relationships. This allows the update device 10 to build a network that links the characteristics of the user's behavioral information on the network with the characteristics of the real-world areas visited by the user.

Next, the update device 10 updates the vectors (user vector and region vector) based on the combined user vector and region vector (step S04). For example, the update device 10 updates the user vector and region vector using GCNs (Graph Convolutional Networks) for bipartite graphs. GCNs is a method of performing convolution on graph data. Convolution using graph data involves adding the feature of one node in the graph multiplied by a weighted feature of an adjacent node.

In other words, the user vector receives the features of adjacent region vectors, and the region vector receives the features of adjacent user vectors and other adjacent region vectors. Therefore, the user vector and the region vector adjacent to the user vector are updated so that they are similar to each other.

In the example shown in the figure, for example, the user vector Vu1 and area vector Va1 are updated with weight w1, the user vector Vu1 and area vector Va2 are updated with weight w2, and the area vector Va1 and area vector Va2 are updated with weight w3.

The weights w1 and w2 between the user vector Vu1 and the area vectors Va1 and Va2 are values that are set based on the frequency and duration of stay of the corresponding users. In other words, if a user stays in area B more frequently than in area A or for a longer period of time, the user vector corresponding to the user will be weighted more heavily for the area vector corresponding to area B than for the area vector corresponding to area A. Furthermore, the weight w3 between area vector Va1 and area vector Va2 is a value that is set, for example, depending on the distance and positional relationship between the corresponding areas. For example, the closer the areas are to each other, the stronger the weight between them.

For example, after updating the user vector and area vector using GCNs, the update device 10 outputs the updated user vector or area vector (step S05). For example, the user vector output from the update device 10 is used for various recommendations on the network, and the user vector output from the update device 10 is used for land development, for example.

In this way, the update device 10 according to the embodiment uses the user's visit history to generate a bipartite graph connecting the user vector and the area vector, and updates the user vector and area vector using the GCNs for the bipartite graph.

For this reason, the update device 10 according to the embodiment can reflect area vectors that indicate the characteristics of areas in the real world on user vectors that indicate the characteristics of behavior on the network, and can reflect user vectors on area vectors.

Therefore, the update device 10 according to the embodiment can appropriately update the vector indicating the features.

2. Example of the configuration of the update device
Next, a configuration example of the update device 10 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing a configuration example of the update device 10 according to the embodiment. As shown in Fig. 2, the update device 10 includes a communication unit 110, a control unit 120, and a storage unit 130.

The communication unit 110 is realized, for example, by a NIC (Network Interface Card). The communication unit 110 transmits and receives information to and from external devices via various wireless communication networks such as 4G (Generation), 5G, LTE (Long Term Evolution), Wi-Fi (registered trademark), or wireless LAN (Local Area Network), or various wired communication networks.

The storage unit 130 is realized, for example, by a semiconductor memory element such as RAM or flash memory, or a storage device such as a hard disk or optical disk. The storage unit 130 also has a user vector model storage unit 131, a user vector storage unit 132, a region vector model storage unit 133, a region vector storage unit 134, a map information storage unit 135, and a bipartite graph storage unit 136.

The user vector model storage unit 131 stores a user vector model. The user vector model is a learning model that converts information about users into user vectors. For example, when information about each user's behavioral history on the network is input into the model, the update device 10 learns the model so that it outputs a user vector similar to the user vectors of other users with similar information.

Here, information related to behavioral history on the network includes, for example, information related to purchase history, search history, browsing history, similarity of objects purchased, searched or viewed on the portal site, and the time (or period) when the purchase, search or viewing took place. Furthermore, the model may be a model that generates user vectors based on information related to each user's usage history on the network as well as information related to user attributes.

Note that the user attributes here may be so-called demographic attributes such as age, gender, residential area, educational background, family structure, place of employment, and annual income, or so-called psychographic attributes such as hobbies, preferences, and lifestyle.

The user vector storage unit 132 stores user vectors. Figure 3 is a diagram showing an example of information stored in the user vector storage unit 132 according to an embodiment. As shown in Figure 3, the user vector storage unit 132 includes information such as "user ID" and "user vector."

The "User ID" field stores identification information for identifying a user. The "User Vector" field stores a user vector corresponding to the user identified by the user ID. Figure 3 shows an example where the user vector is two-dimensional vector data "10, 24", but the user vector can be multi-dimensional (N-dimensional) vector data and is not limited to two dimensions.

The region vector model storage unit 133 stores region vector models. Region vector models are learning models that convert region-related information into region vectors. For example, when region-related information is input to the model, the update device 10 learns the model so that it outputs a user vector similar to the user vectors of other users with similar information.

Note that the information about the area here includes information about the facilities located there, roads, and blocks. The area vector model may also be a model that generates an area vector by averaging the user vectors of each user who visits.

The region vector storage unit 134 stores region vectors. Figure 4 is a diagram showing an example of the region vector storage unit 134 according to an embodiment. The region vector storage unit 134 contains information such as "region ID" and "region vector."

The "Area ID" field stores identification information for identifying the area. The "Area Vector" field stores the area vector of the user identified by the area ID. Figure 4 shows an example where the area vector is two-dimensional vector data "33, 15", but the user vector can be multi-dimensional (N-dimensional) vector data and is not limited to two dimensions.

The map information storage unit 135 stores map information that indicates the positional relationships between each region. For example, weights of region vectors between regions are set based on this map information.

The bipartite graph storage unit 136 stores the bipartite graph shown in Figure 1. As shown in Figure 1, the bipartite graph is a graph that includes a set of users and a set of areas. In the bipartite graph, each user and each area is a node, and edges connecting users and areas are set based on visit history, while edges connecting areas are set based on the geographic relationships between the areas.

The control unit 120 is, for example, a controller, and is realized by a CPU (Central Processing Unit) or MPU (Micro Processing Unit) executing various programs stored in the storage device inside the update device 10 using RAM as a working area. The control unit 120 is also a controller, and is realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

As shown in FIG. 2, the control unit 120 has a user vector generation unit 121, a region vector generation unit 122, a combining unit 123, a setting unit 124, an update unit 125, and a providing unit 126.

The user vector generation unit 121 generates a user vector that indicates the characteristics of a user based on the user's network behavior. For example, the user vector generation unit 121 periodically acquires information about the user's network behavior from a portal site or the user terminal owned by each user.

The user vector generation unit 121 generates a user vector by inputting the acquired information about the user's network behavior into the user vector model stored in the user vector model storage unit 131. The user vector generation unit 121 also stores the generated user vector in the user vector storage unit 132.

The area vector generation unit 122 generates area vectors based on information about the facilities, roads, and sections of the area. For example, the area vector generation unit 122 obtains current map information for each area from an external device such as a server, and generates area vectors by inputting the obtained map information into the area vector model stored in the area vector model storage unit 133.

The area vector generation unit 122 may generate area vectors based on the user vectors of users who visited the area. In this case, for example, the area vector generation unit 122 may weight the user vectors of each user for each area based on the user's visit frequency, visit duration, etc., and then generate an area vector by averaging these values. The area vector generation unit 122 may also generate one area vector for each area using the map information and the user vectors.

The combining unit 123 combines nodes corresponding to users with nodes corresponding to areas in the bipartite graph based on the user's visit history. For example, the combining unit 123 periodically acquires location information (e.g., GPS positioning information) of each user terminal and calculates the visit history of each user to each area.

Next, the combining unit 123 combines the node corresponding to the user with the node corresponding to the area visited by the user based on the calculated visit history. The combining unit 123 also combines the areas according to the geographical relationship between the areas. In other words, the combining unit 123 combines areas by combining each area with an adjacent area.

The setting unit 124 sets the weight for updates between users and areas according to the relationship between the user and area based on the visit history. Here, the weight for updates refers to the degree of influence that a user vector (or area vector) has on an adjacent area vector (or user vector). When updating vectors, the stronger the weight for updates, the more strongly both vectors are influenced by each other's vectors.

For example, the setting unit 124 learns the relationship between users and areas based on each user's visit history, and sets a weight for updates between users and areas according to the learned relationship.

The relationship between a user and an area here refers to the connection between the user and the area in the real world; for example, the more frequently a user visits a particular area or the longer the visit, the stronger the connection between the two.

To this end, the setting unit 124 analyzes the visit history using a predetermined algorithm and sets weights for updates between users and areas.

Furthermore, for example, the setting unit 124 may set the weighting for updates between users and areas so that it gradually weakens over time. In other words, in this case, the more recent the date and time of the visit, the stronger the weighting, and the older the date and time of the visit, the weaker the weighting is set.

The combining unit 123 may also use either a valid edge or a invalid edge for the edge connecting the node corresponding to the user with the node corresponding to the area. For example, if a valid edge is used, it becomes possible to design a bipartite graph in which the user vector affects the area vector, but the area vector does not affect the user vector.

The update unit 125 updates the user vector or area vector based on a user vector that indicates the characteristics of the user based on the user's behavior on the network and an area vector that indicates the characteristics of the area combined with the user by the combination unit 123.

For example, the update unit 125 updates each user vector and each region vector using the GCNs for the bipartite graph stored in the bipartite graph storage unit 136. In this case, the update unit 125 may update only one of the user vector and the region vector.

That is, for example, the update unit 125 may update the region vector based on the user vector, or may update the user vector based on the region vector. Furthermore, the update timing and update period of the user vector may be set separately from the update timing and update period of the region vector.

The update unit 125 also updates each vector according to the update weight set by the setting unit 124. The stronger the update weight, the greater the influence both vectors have on each other, and the weaker the update weight, the less likely the two vectors are to influence each other.

Here, an example of the update process will be described using Figure 5. Figure 5 is a schematic diagram of a user vector and a region vector according to an embodiment. In the example shown in Figure 5(a), the user vector Vu and the region vector Va are vectors of the same dimension, and the update unit 125 updates the user vector Vu or the region vector Va by multiplying the user vector Vu (or the region vector Va) to be updated by the region vector Va (or the user vector Vu) that is adjacent to it.

In this case, for example, the data for each dimension in the user vector Vu will be influenced by the characteristics of the adjacent region vector Va, and the data for each dimension in the region vector Va will be influenced by the characteristics of the adjacent user vector Vu.

On the other hand, in the example shown in Figure 5(b), for example, the user vector Vu and the area vector Va are set in separate dimensions to prevent them from being mixed, and then the user vector Vu or area vector Va is updated.

In this case, if a user vector having a user vector Vu1 and a region vector Va1 that is blank data is updated using a region vector having a user vector Vu2 that is blank data and a region vector Va2, the user vector will be updated to have a user vector Vu1 and a region vector Va2.

In other words, in this case, the area vector area of the user vector reflects the characteristics of the area visited, and the user vector area of the area vector reflects the characteristics of the visiting user.

The update unit 125 may also update the area vectors of each area using the area vectors of areas adjacent to each area at a predetermined interval (for example, once a day). Figure 6 is a schematic diagram showing an example of updating area vectors according to an embodiment.

As shown in Figure 6, each region corresponds to a piece of land in the real world, so each node has a geographical relationship with the others. Therefore, for example, each region vector is updated so that geographically close regions have similar characteristics.

For example, as shown in Figure 6, among the area vectors Va11 to Va33 corresponding to each of the nine divided areas, the central area vector Va22 is updated by the adjacent area vectors Va11 to Va21 and area vectors Va23 to Va33.

In other words, the update unit 125 updates the area vector Va22 so that it is similar to each of the adjacent area vectors. This allows the area vectors to be updated appropriately.

For example, when updating the region vectors of each region, the update unit 125 may generate a graph in which adjacent nodes are connected by edges, and then update each region vector using the GCNs for this graph.

Returning to the explanation of Figure 2, the providing unit 126 will now be described. The providing unit 126 provides the user vector or region vector updated by the updating unit 125. For example, the providing unit 126 provides the user vector to a portal site, which can then provide various services that utilize the user vector to users.

Furthermore, for example, the providing unit 126 can provide area vectors to facilities such as shops in the corresponding area, enabling the facilities to develop services that take into account the characteristics of the area.

3. Processing Procedure
Next, a processing procedure executed by the update device 10 according to the embodiment will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of update processing according to the embodiment.

As shown in FIG. 7, the update device 10 first sets up a bipartite graph in which users and areas are nodes (step S101). Next, the update device combines nodes (users and areas) based on the user's visit history (step S102).

Next, the update device 10 sets weights for the edges between each node (step S103). Then, the update device 10 updates the vector (user vector or region vector) (step S104) and ends the process.

4. Modifications
In the above embodiment, a user vector based on the user's online behavior and a region vector that indicates the characteristics of a region in the real world are combined based on the visit history. However, the region can be expanded to various targets, for example, by combining a user with a product that the user purchased in the real world and then updating the user vector and the product vector that indicates the characteristics of the product.

5. Effects
The update device 10 according to the embodiment described above includes a combining unit 123 that combines a user with an area visited by the user based on the user's visit history in a bipartite graph including a set of users and a set of areas, and an update unit 125 that updates a user vector or an area vector based on a user vector indicating a characteristic of the user based on the user's behavior on the network and an area vector indicating a characteristic of the area combined with the user by the combining unit 123. Therefore, the update device 10 according to the embodiment can update a user vector based on the area vector and can update an area vector based on the user vector, and therefore can appropriately update a vector indicating a characteristic.

Furthermore, the combining unit 123 according to the embodiment combines regions according to the geographical relationship between the regions, and the updating unit 125 updates the region vectors of the regions using weights according to the geographical relationship. Therefore, the updating device 10 according to the embodiment can appropriately update each region vector according to the geographical positional relationship.

In addition, the update unit 125 according to the embodiment updates the user vector or region vector so that the user vector and region vector corresponding to adjacent users and regions, respectively, in the bipartite graph are similar to each other. Therefore, the update device 10 according to the embodiment can appropriately update each vector.

The update device 10 according to the embodiment also includes a setting unit 124 that sets a weight for updates between users and areas according to the relationship between the users and areas based on their visit history, and the update unit 125 updates the user vector or area vector according to the weight for updates set by the setting unit 124. Therefore, the update device 10 according to the embodiment can appropriately set weights in the bipartite graph, thereby appropriately updating each vector.

Furthermore, the setting unit 124 according to the embodiment sets the weight for updates based on the frequency of visits to an area by a user and the length of time spent there. Therefore, the update device 10 according to the embodiment can appropriately set the weight for updates.

Furthermore, the setting unit 124 according to the embodiment lowers the weighting for updates between a user and an area the older the user's visit. Therefore, the update device 10 according to the embodiment can appropriately update the bipartite graph, thereby appropriately updating each vector.

The update device 10 according to the embodiment also includes a user vector generation unit 121 that generates a user vector for a user based on the user's purchase history, search history, and browsing history on the network. Therefore, the update device 10 according to the embodiment can generate a user vector that captures the user's network characteristics.

In addition, the user vector generation unit 121 according to the embodiment generates a user vector based on information about the user's user attributes. Therefore, the update device 10 according to the embodiment can generate a more appropriate user vector.

The update device 10 according to the embodiment also includes a region vector generation unit 122 that generates a region vector based on the user vectors of users who have visited the region. Therefore, the update device 10 according to the embodiment can generate a region vector that takes into account the user vectors of users who have visited the region.

In addition, the region vector generation unit 122 according to the embodiment generates region vectors based on information about the facilities, roads, and sections of the region. Therefore, the update device 10 according to the embodiment can generate more similar region vectors for regions with more similar facilities, etc.

An update method according to an embodiment is a computer-executed update method that includes a combining step of combining a user with an area visited by the user based on the user's visit history in a bipartite graph including a set of users and a set of areas, and an update step of updating a user vector or area vector based on a user vector that indicates the characteristics of the user based on the user's behavior on the network and an area vector that indicates the characteristics of the area combined with the user in the combining step. Therefore, the update method according to an embodiment can appropriately update the vector indicating the characteristics.

The update program according to the embodiment also causes a computer to execute a combination procedure in a bipartite graph including a set of users and a set of areas, in which a user is combined with an area visited by the user based on the user's visit history, and an update procedure in which a user vector indicating the characteristics of the user based on the user's behavior on the network and an area vector indicating the characteristics of the area combined with the user by the combination procedure are updated. Therefore, the update program according to the embodiment can appropriately update the vector indicating the characteristics.

6. Hardware Configuration
The update device 10 according to the embodiment described above is realized by a computer 1000 having a configuration as shown in Fig. 8. Fig. 8 is a hardware configuration diagram showing an example of a computer that realizes the functions of the update device 10 according to the embodiment. The computer 1000 has a CPU 1100, a RAM 1200, a ROM 1300, a HDD 1400, a communication interface (I/F) 1500, an input/output interface (I/F) 1600, and a media interface (I/F) 1700.

The CPU 1100 operates based on programs stored in the ROM 1300 or HDD 1400, and controls each component. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 starts up, as well as programs that depend on the computer 1000's hardware.

HDD 1400 stores programs executed by CPU 1100 and data used by such programs. Communication interface 1500 receives data from other devices via network (communication network) N and sends it to CPU 1100, and transmits data generated by CPU 1100 to other devices via network N.

The CPU 1100 controls output devices such as displays and printers, and input devices such as keyboards and mice (in Figure 8, the output devices and input devices are collectively referred to as "input/output devices") via the input/output interface 1600. The CPU 1100 acquires data from the input devices via the input/output interface 1600. The CPU 1100 also outputs generated data to the output devices via the input/output interface 1600.

Media interface 1700 reads programs or data stored on recording medium 1800 and provides them to CPU 1100 via RAM 1200. CPU 1100 loads the programs from recording medium 1800 onto RAM 1200 via media interface 1700 and executes the loaded programs. Recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or a PD (Phase Change Rewritable Disc), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory.

For example, when the computer 1000 functions as the update device 10 according to the embodiment, the CPU 1100 of the computer 1000 executes programs loaded onto the RAM 1200, thereby realizing the functions of the control unit 120. The CPU 1100 of the computer 1000 reads and executes these programs from the recording medium 1800, but as another example, the CPU 1100 may also obtain these programs from another device via the network N.

The above describes in detail some of the embodiments of the present application based on the drawings, but these are merely examples, and the present invention can be implemented in other forms that incorporate various modifications and improvements based on the knowledge of those skilled in the art, including the aspects described in the Disclosure of the Invention section.

[7. Other]
Furthermore, among the processes described in the above embodiments and modifications, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, the various information shown in each drawing is not limited to the information shown in the drawings.

Furthermore, the components of each device shown in the figure are functional concepts and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

Furthermore, the above-described embodiments and variations can be combined as appropriate to the extent that the processing content is not contradictory.

Furthermore, the "section, module, unit" mentioned above can be read as "means" or "circuit." For example, an acquisition unit can be read as an acquisition means or an acquisition circuit.

10 Update device 110 Communication unit 120 Control unit 121 User vector generation unit 122 Area vector generation unit 123 Combination unit 124 Setting unit 125 Update unit 126 Provision unit 130 Storage unit 131 User vector model storage unit 132 User vector storage unit 133 Area vector model storage unit 134 Area vector storage unit 135 Map information storage unit 136 Bipartite graph storage unit Va Area vector Vu User vector

Claims (12)

a linking unit that links a user and the area visited by the user based on the user's visit history in the real world in a bipartite graph including a set of users and a set of areas;
and an update unit that updates the user vector or the region vector based on a user vector that indicates characteristics of the user based on the user's behavior on the network and a region vector that indicates characteristics of the region that has been combined with the user by the combination unit. The coupling portion is
combining the regions according to a geographical relationship between the regions;
The update unit
The updating device according to claim 1 , wherein the area vectors of the areas are updated using weights according to the geographical relationships. The update unit
The update device according to claim 1 or 2, characterized in that the user vector or the region vector corresponding to the user and the region adjacent to each other in the bipartite graph are updated so that the user vector and the region vector are similar to each other. a setting unit that sets a weight related to updates between the user and the area in accordance with a relationship between the user and the area based on the visit history;
The update unit
The updating device according to claim 1 , 2 or 3 , wherein the user vector or the region vector is updated in accordance with the weight related to the update set by the setting unit. The setting unit
5. The update device according to claim 4, wherein a weight for the update is set based on the frequency of visits and duration of stay of the user in the area. The setting unit
6. The updating device according to claim 4, wherein the older the time of the user's visit, the lower the weighting of the update between the user and the area. The update device according to any one of claims 1 to 6, further comprising a user vector generation unit that generates the user vector of the user based on the user's purchase history, search history, and browsing history on the network. The user vector generation unit
The updating device according to claim 7 , wherein the user vector is generated based on information relating to user attributes of the user. 9. The update device according to claim 1, further comprising: a region vector generation unit that generates the region vector based on the user vector of the user who visited the region. The area vector generation unit
The updating device according to claim 9, wherein the area vector is generated based on information about facilities, roads, and sections of the area. 1. A computer-implemented update method, comprising:
a linking step of linking a user with the area visited by the user based on the user's visit history in the real world in a bipartite graph including a set of users and a set of areas;
an updating step of updating the user vector or the area vector based on a user vector indicating the characteristics of the user based on the user's behavior on the network and an area vector indicating the characteristics of the area combined with the user by the combining step. a linking procedure for linking a user with the areas visited by the user based on the user's visit history in the real world in a bipartite graph including a set of users and a set of areas;
and an update procedure for updating the user vector or the area vector based on a user vector indicating characteristics of the user based on the user's behavior on the network and an area vector indicating characteristics of the area combined with the user by the combining procedure.