Information processing system, information processing apparatus, information processing method, and program

Abstract

To enhance anonymity of information to be shared.SOLUTION: An information processing system includes an information processing apparatus, a first terminal device that holds first user characteristic information, and a second terminal device that holds second user characteristic information. The information processing apparatus includes: an information collection unit which collects, from the second terminal device, anonymized information obtained by performing predetermined anonymization processing on the second user characteristic information; a learning unit which generates a model configured to learn, by machine learning, a relationship between the first user characteristic information and the anonymized information, using the first user characteristic information and the anonymized information, as learning data, and output, on receipt of the first user characteristic information, estimated user characteristic information estimated from the relationship between the first user characteristic information and the anonymized information; and a model output unit which outputs the model to the first terminal device.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information processing system, an information processing device, an information processing method, and a program. [Background technology]

[0002] In recent years, the revised Personal Information Protection Act has defined the new concept of anonymously processed information and established rules for safely sharing information between companies. In line with this, in recent years, information sharing between businesses has become common in information sharing environments (data clean rooms) that protect the privacy of information and ensure security while sharing information between different companies. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2020-507826 Summary of the Invention [Problem to be solved by the invention]

[0004] In a conventional data sharing environment, if the business providing the information does not restrict how the information is handled by the recipient (the business using the information), the anonymity of the information may not be guaranteed.

[0005] The disclosed technology aims to increase the anonymity of shared information. [Means for solving the problem]

[0006] The disclosed technology is an information processing system having an information processing device, a first terminal device that stores first user characteristic information, and a second terminal device that stores second user characteristic information, wherein the information processing device has: an information collection unit that collects, from the second terminal device, anonymized processed information obtained by performing a predetermined anonymization process on the second user characteristic information; a learning unit that uses the first user characteristic information and the anonymized processed information as learning data to perform machine learning on the relationship between the first user characteristic information and the anonymized processed information, and that generates a model that outputs estimated user characteristic information estimated from the relationship between the first user characteristic information and the anonymized processed information when the first user characteristic information is input; and a model output unit that outputs the model to the first terminal device. [Effects of the Invention]

[0007] The anonymity of the information shared can be increased. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 illustrates an example of a system configuration of an information processing system. [Figure 2] FIG. 2 illustrates an example of a hardware configuration of an information processing device. [Figure 3] FIG. 2 is a diagram illustrating the functional configuration of each device included in the information processing system. [Figure 4] FIG. 10 is a diagram illustrating an example of a user characteristic information storage unit of a user terminal. [Figure 5] FIG. 10 is a diagram illustrating an example of a store characteristic information storage unit of a user terminal. [Figure 6] FIG. 10 is a diagram illustrating an example of a product feature information storage unit of a user terminal. [Figure 7] FIG. 10 is a diagram illustrating an example of an association information storage unit of a user terminal. [Figure 8] FIG. 2 is a sequence diagram illustrating the operation of the information processing system. [Figure 9] 10 is a flowchart illustrating an information processing apparatus. [Figure 10] FIG. 10 is a diagram illustrating the processing of a model generation unit. [Figure 11] 10 is a first flowchart illustrating processing of a terminal device. [Figure 12] 10 is a second flowchart illustrating the processing of the terminal device. [Figure 13] FIG. 10 is a first diagram showing a display example of a terminal device. [Figure 14] FIG. 10 is a second diagram showing an example of a display on the terminal device. DETAILED DESCRIPTION OF THE INVENTION

[0009] The present embodiment will be described below with reference to the drawings: Fig. 1 is a diagram showing an example of the system configuration of an information processing system.

[0010] The information processing system 100 of this embodiment includes an information processing device 200, a terminal device 300, and a terminal device 400, and the terminal device 300 and the terminal device 400 are each connected to the information processing device 200 via a network such as the Internet.

[0011] The terminal device 300 and the terminal device 400 of this embodiment may be terminals managed by different companies.

[0012] In this embodiment, the information processing device 200 and the terminal device 300 are managed by the same business operator. The business operator that manages the information processing device 200 and the terminal device 300 is a business operator that uses information provided by the business operator that manages the terminal device 400. In the following description, the business operator that manages the terminal device 300 may be referred to as a user.

[0013] Furthermore, the terminal device 300 in this embodiment is a user terminal that is mainly used to utilize information provided by the business operator that manages the terminal device 400, and the terminal device 400 is mainly a provider terminal that provides information to other businesses.

[0014] The information processing device 200 of this embodiment is managed by a user (a business that manages the terminal device 300), and stores characteristic information provided by a provider. In other words, the information processing device 200 of this embodiment realizes a data clean room that allows users to share data provided by providers. A data clean room refers to a data sharing environment in which data can be shared between different companies while protecting data privacy and ensuring security.

[0015] The business operator that manages the terminal device 300 of this embodiment holds characteristic information for each of a plurality of types of entities within the company, and the characteristic information for each of the plurality of types of entities is stored in the terminal device 300. The characteristic information stored in the terminal device 300 is an example of first characteristic information.

[0016] The business operator that manages the terminal device 400 of this embodiment holds characteristic information for each of a plurality of types of entities within the company, and the characteristic information for each of the plurality of types of entities is stored in the terminal device 400. The characteristic information stored in the terminal device 400 is an example of second characteristic information.

[0017] Furthermore, the terminal device 300 of this embodiment stores linking information linking multiple types of entities together, and maps the multiple types of entities onto a single vector space using the respective feature information of the multiple types of entities and the linking information.The terminal device 300 of this embodiment then compresses the vector space onto which the multiple types of entities are mapped into a low-dimensional space and displays it on the screen.

[0018] Therefore, according to this embodiment, it is possible for the business operator who uses the terminal device 300 to allow the user of the terminal device 300 to visually understand the relationships between multiple types of entities.

[0019] Furthermore, the information processing device 200 of this embodiment generates a model that learns the relationship between the feature information held by the terminal device 300 and the anonymously processed information from the terminal device 400. Then, the information processing device 200 provides the generated model to the terminal device 300.

[0020] The terminal device 300 of this embodiment inputs feature information stored in the terminal device 300 into the model provided from the information processing device 200, and obtains estimated feature information estimated by the model. The estimated feature information is anonymized information that is estimated to be most highly related to the feature information input to the model, among the anonymized information provided from the terminal device 400.

[0021] In this way, the terminal device 300 of this embodiment can use the feature information provided by the provider to obtain estimated feature information that is highly relevant to the feature information stored in the terminal device 300. Furthermore, the terminal device 300 of this embodiment also uses the estimated feature information when mapping multiple types of entities onto a vector space. Therefore, according to this embodiment, the operator of the terminal device 300 can expand and utilize the feature information that they own based on the information provided by the provider.

[0022] 1, the information processing system 100 includes one information processing device 200 and one terminal device 300, 400, but this is not limited to this. The information processing device 200 of this embodiment may be realized by a plurality of information processing devices. Furthermore, the information processing system 100 of this embodiment may include any number of terminal devices.

[0023] Next, the hardware configuration of the information processing device 200 of this embodiment will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the hardware configuration of the information processing device.

[0024] The information processing device 200 of this embodiment includes a central processing unit (CPU) 201, a read-only memory (ROM) 202, and a random access memory (RAM) 203. The CPU 201, the ROM 202, and the RAM 203 form a so-called computer.

[0025] The information processing device 200 also includes an auxiliary storage device 204 , a display device 205 , an operation device 206 , an I / F (Interface) device 207 , and a drive device 208 .

[0026] The hardware components of the information processing device 200 are connected to one another via a bus B.

[0027] The CPU 201 is a computing device that executes various programs installed in the auxiliary storage device 204 .

[0028] The ROM 202 is a non-volatile memory. The ROM 202 functions as a main storage device that stores various programs, data, etc. required for the CPU 201 to execute various programs installed in the auxiliary storage device 204. Specifically, the ROM 202 functions as a main storage device that stores boot programs such as a BIOS (Basic Input / Output System) and an EFI (Extensible Firmware Interface).

[0029] The RAM 203 is a volatile memory such as a dynamic random access memory (DRAM) or a static random access memory (SRAM). The RAM 203 functions as a main storage device that provides a working area in which various programs installed in the auxiliary storage device 204 are expanded when the CPU 201 executes them.

[0030] The auxiliary storage device 204 is an auxiliary storage device that stores various programs and information used when the various programs are executed.

[0031] The display device 205 is a display device that displays the internal state of the information processing device 200. The operation device 206 is an input device that allows the person operating the information processing device 200 to input various instructions to the information processing device 200.

[0032] The I / F device 207 is a communication device that connects to a network and communicates with other devices.

[0033] The drive device 208 is a device for loading a recording medium 209. The recording medium 209 here includes media that record information optically, electrically, or magnetically, such as a CD-ROM, a flexible disk, or a magneto-optical disk. The recording medium 209 may also include semiconductor memories that record information electrically, such as an EPROM (Erasable Programmable Read Only Memory) or a flash memory.

[0034] The various programs to be installed in the auxiliary storage device 204 are installed, for example, by setting the distributed recording medium 209 in the drive device 208 and reading the various programs recorded on the recording medium 209 by the drive device 208. Alternatively, the various programs to be installed in the auxiliary storage device 204 may be installed by being downloaded from a network via the I / F device 207.

[0035] Furthermore, the terminal devices 300 and 400 of this embodiment may be, for example, tablet-type terminal devices, and the hardware configuration may be the same as that of the information processing device 200.

[0036] Next, the functional configuration of each device included in the information processing system 100 of this embodiment will be described with reference to FIG.

[0037] 3 is a diagram illustrating the functional configuration of each device in the information processing system. First, the functions of the information processing device 200 will be described.

[0038] In the following description of this embodiment, the characteristic information provided from the terminal device 400 to the information processing device 200 is assumed to be anonymously processed information obtained by applying a predetermined anonymization process (appropriate processing) to user characteristic information held by the business operator that manages the terminal device 400.

[0039] The information processing device 200 of this embodiment includes an anonymously processed information storage unit 210 , a model storage unit 230 , an information collection unit 250 , and a model generation unit 260 .

[0040] The anonymously processed information storage unit 210 and the model storage unit 230 may be realized by the auxiliary storage device 204 or the like of the information processing device 200. Furthermore, the information collection unit 250 and the model generation unit 260 may be realized by the CPU 201 of the information processing device 200 reading and executing a program stored in the ROM 202, the RAM 203, or the like.

[0041] Anonymously processed information is stored in the anonymously processed information storage unit 210. The anonymously processed information is user characteristic information that has been subjected to a predetermined anonymization process in the terminal device 400 (provider terminal).

[0042] The model storage unit 230 stores the model generated by the model generation unit 260.

[0043] The information collecting unit 250 collects anonymously processed information from a provider terminal included in the information processing system 100. More specifically, the information collecting unit 250 collects anonymously processed information, which is user characteristic information anonymized in the terminal device 400, from the terminal device 400. Note that the information processing system 100 may include a plurality of terminal devices 400, and the information collecting unit 250 may collect anonymously processed information from a plurality of provider terminals.

[0044] The model generation unit 260 receives user characteristic information of a user as input and generates a model that estimates and outputs anonymously processed information regarding each piece of user characteristic information. In other words, the model generation unit 260 receives user characteristic information stored in the terminal device 300 as input and generates a model that estimates and outputs anonymously processed information regarding each piece of user characteristic information. In the following description, the model generated by the model generation unit 260 may be referred to as an estimated model.

[0045] The model generation unit 260 of this embodiment includes a training data generation unit 262 , a training unit 263 , and a model output unit 264 .

[0046] The learning data generation unit 262 generates learning data for generating an estimation model based on the anonymously processed information.

[0047] The learning unit 263 performs machine learning on the learning data to generate an estimation model. The generated estimation model is stored in the model storage unit 230. The model output unit 264 outputs the estimation model 326 generated by the learning unit 263 to the terminal device 300.

[0048] In this embodiment, the anonymously processed information stored in the anonymously processed information storage unit 210 is If "public disclosure" as stipulated in the Personal Information Protection Act has been made, provision of the information to third parties may be permitted.

[0049] Next, we will explain the functions of the terminal device 300. The terminal device 300 of this embodiment includes a user characteristic information storage unit 305, a store characteristic information storage unit 306, a product characteristic information storage unit 307, a linking information storage unit 308, a visualization processing unit 320, a list generation unit 330, a distribution processing unit 340, and a demand forecasting unit 350.

[0050] The user characteristic information storage unit 305 stores user characteristic information indicating the characteristics of a user. The user characteristic information stored in the user characteristic information storage unit 305 is an example of first user characteristic information. The store characteristic information storage unit 306 stores store characteristic information indicating the characteristics of a store. The product characteristic information storage unit 307 stores product characteristic information indicating the characteristic information of a product. Each of the user characteristic information, store characteristic information, and product characteristic information is an example of characteristic information indicating the characteristics of an entity.

[0051] The linking information storage unit 308 stores log information indicating the purchase history when a user purchases a product at a store. The log information in this embodiment includes information that identifies the user, information that identifies the store, and information that identifies the product. The log information in this embodiment is an example of linking information that links user characteristic information, store characteristic information, and product characteristic information.

[0052] The user characteristic information, store characteristic information, product characteristic information, and log information stored in each memory unit of the terminal device 300 in this embodiment is information held by the business operator that manages the terminal device 300, and in this embodiment, is not shared with other businesses.

[0053] In addition, in the present embodiment, the user characteristic information storage unit 305, the store characteristic information storage unit 306, the product characteristic information storage unit 307, and the linking information storage unit 308 are included in the terminal device 300, but this is not limiting. The user characteristic information storage unit 305, the store characteristic information storage unit 306, the product characteristic information storage unit 307, and the linking information storage unit 308 may be provided in the information processing device 200.

[0054] The visualization processing unit 320 uses the user feature information, store feature information, product feature information, and log information to map the relationships between users, stores, and products into a single vector space, and displays the results on the terminal device 300. In other words, the visualization processing unit 320 of this embodiment displays on the terminal device 300 a screen that allows the user to visually grasp the relationships between multiple types of entities.

[0055] Furthermore, when performing mapping, the visualization processing unit 320 of this embodiment may include estimated user feature information estimated by the estimation model as part of the user feature information. The estimated user feature information is anonymized information that is estimated to be most highly related to the user feature information input to the estimation model 326, among the anonymously processed information provided from the terminal device 400.

[0056] The visualization processing unit 320 includes an input receiving unit 321 , an information expanding unit 322 , a vector conversion unit 323 , a group generating unit 324 , and a screen generating unit 325 .

[0057] The input receiving unit 321 receives various inputs to the terminal device 300. The information expanding unit 322 holds an estimation model 326 transmitted from the information processing device 200, and acquires estimated user characteristic information using the estimation model 326. Details of the processing by the information expanding unit 322 will be described later.

[0058] The vector conversion unit 323 converts each of the users, stores, and products into a vector representation based on the user feature information including estimated user feature information, store feature information, product feature information, and log information, and represents them in a single vector space. In other words, the vector conversion unit 323 maps multiple types of entities onto a single vector space based on linking information that links the feature information of each entity.

[0059] More specifically, the vector conversion unit 323 maps multiple types of entities onto a vector space learned by one of the following methods.

[0060] Method 1: Item2Vec: Neural Item Embedding for Collaborative Filtering Method 2: Billion-scale Commodity Embedding for E-commerce Recommendation in Alibaba Method 3: DTCDR: A Framework for Dual-Target Cross-Domain Recommendation Method 4: Cross-Domain Recommendation: An Embedding and Mapping Approach Note that the method of mapping multiple types of entities onto a vector space may be a method other than the four methods described above.

[0061] In this embodiment, when an entity is selected in a vector space in which multiple types of entities are mapped, the group generation unit 324 creates a group according to the selected entity. Furthermore, when a search condition for feature information is input, the group generation unit 324 may group entities identified from the search results based on the search condition into one group.

[0062] The screen generation unit 325 compresses the vector space onto which the multiple types of entities are mapped by the vector conversion unit 323 into a low-dimensional space (two or three dimensions), and generates screen data to be displayed on the terminal device 300.

[0063] Furthermore, the screen generation unit 325 controls the display of the terminal device 300 in response to various operations on the terminal device 300. In other words, the screen generation unit 325 is an example of a display control unit.

[0064] The list generation unit 330 generates a list of users to which advertising information is to be distributed. The distribution processing unit 340 distributes the advertising information to users based on the list generated by the list generation unit 330. The demand prediction unit 350 predicts demand for products.

[0065] Next, a description will be given of the functional configuration of the terminal device 400. The terminal device 400 includes a user characteristic information storage unit 405, a store characteristic information storage unit 406, a product characteristic information storage unit 407, a linking information storage unit 408, and an anonymization processing unit 410.

[0066] The user characteristic information storage unit 405 stores user characteristic information indicating the characteristics of a user. The user characteristic information stored in the user characteristic information storage unit 405 is an example of second user characteristic information. The store characteristic information storage unit 406 stores store characteristic information indicating the characteristics of a store. The product characteristic information storage unit 407 stores product characteristic information indicating the characteristic information of a product. Each of the user characteristic information, store characteristic information, and product characteristic information is an example of characteristic information indicating the characteristics of an entity.

[0067] The linking information storage unit 408 stores log information indicating the purchase history when a user purchases a product at a store. The log information in this embodiment includes information that identifies the user, information that identifies the store, and information that identifies the product. The log information in this embodiment is also an example of linking information that links user characteristic information, store characteristic information, and product characteristic information.

[0068] The user characteristic information, store characteristic information, product characteristic information, and log information stored in each storage unit of the terminal device 400 of this embodiment are information held by the business operator that manages the terminal device 400.

[0069] The anonymization processing unit 410 performs a predetermined anonymization processing on the user characteristic information stored in the user characteristic information storage unit 405 , converts the user characteristic information into anonymous processed information, and transmits the anonymous processed information to the information processing device 200 .

[0070] The anonymously processed information transmitted to the information processing device 200 is stored in the anonymously processed information storage unit 210 and provided to the business operator that manages the terminal device 300 via the information processing device 200.

[0071] In addition, the terminal device 400 of this embodiment may perform anonymization processing on the store feature information stored in the store feature information storage unit 406 and the product feature information stored in the product feature information storage unit 407 using the anonymization processing unit 410.

[0072] Here, the anonymization processing performed by the anonymization processing unit 410 of the terminal device 400 will be described. Note that the anonymization processing unit 410 of this embodiment may be realized by an application installed in the terminal device 400. Furthermore, the application that realizes the anonymization processing unit 410 may be distributed from the information processing device 200 to the terminal device 400.

[0073] In this embodiment, k-anonymity, which is a probabilistic index, is introduced as the predetermined anonymization process. Specifically, the anonymization processing unit 410 of this embodiment defines quasi-identifiers, which are attributes that can identify an individual by combining them rather than by themselves, and performs processing to generalize user characteristic information or delete each record so that there are k or more pieces of data with the same attributes.

[0074] Although various researches have been conducted on k-anonymization techniques, in this embodiment, a k-anonymization technique that generalizes user feature information without reducing the amount of information too much is selected. By selecting such a technique, highly accurate analysis can be performed.

[0075] Furthermore, in order to enhance confidentiality, the anonymization processing unit 410 of this embodiment may add noise to the user feature information anonymized based on k-anonymity to an extent that satisfies differential privacy, to obtain anonymously processed information. Note that adding noise may also be performed on feature information other than user feature information. For example, noise may be added to a unit price included in product feature information.

[0076] In this embodiment, this reduces privacy risks when the estimation model 326 generated using anonymously processed information predicts the input data used for learning in the same form as before.

[0077] Furthermore, in this embodiment, by making the user characteristic information provided to other businesses into anonymously processed information, it is possible to reduce the risk that individuals will be guessed at the recipient of the information.

[0078] It is assumed that the probabilistic index k, the value of the privacy level ε, the information items to be used as quasi-identifiers, etc. are determined in advance through agreement between the information provider and the information user.

[0079] In this embodiment, noise is added to the user feature information anonymized based on k-anonymity to a degree that satisfies differential privacy, but this is not limiting. In this embodiment, the user feature information anonymized based on k-anonymity may be transmitted to the information processing device 200 as anonymously processed information.

[0080] Next, each storage unit included in the terminal device 300 will be described with reference to FIGS.

[0081] FIG. 4 is a diagram showing an example of a user characteristic information storage unit of an end user terminal. The user characteristic information stored in the user characteristic information storage unit 305 of this embodiment includes information items such as user ID, age, sex, address, and name. The user characteristic information is information including the value of the item "user ID" and values ​​of other items. In this embodiment, the user characteristic information may be stored in advance in the user characteristic information storage unit 305 of the terminal device 300 by an administrator of the terminal device 300 or the like.

[0082] In the user characteristic information, the value of the item "User ID" is identification information for identifying the user, and the values ​​of the items "Age," "Gender," and "Address" indicate the user's age, gender, and address. Note that the information items included in the user characteristic information are not limited to the items shown in Fig. 4, and may include items other than those shown in Fig. 4, or may not include all of the items shown in Fig. 4.

[0083] The user characteristic information in this embodiment may include, for example, the time period during which the user identified by the user ID purchases products, the type of products purchased, the frequency, and the like.

[0084] 5 is a diagram showing an example of a store characteristic information storage unit of a user terminal. The store characteristic information stored in the store characteristic information storage unit 306 of this embodiment includes information items such as store ID, type, and address. The store characteristic information is information including the value of the item "store ID" and values ​​of other items. In this embodiment, the store characteristic information may be stored in advance in the store characteristic information storage unit 306 of the terminal device 300 by an administrator of the terminal device 300 or the like.

[0085] In the store characteristic information, the value of the item "Store ID" is identification information for identifying a store, and the value of the item "Type" indicates the type of store. Store types may include, for example, clothing stores, restaurants, supermarkets, etc. The value of the item "Address" indicates the address of the store. Note that the information items included in the store characteristic information are not limited to the items shown in FIG. 5, and may include items other than those shown in FIG. 5, or may not include all of the items shown in FIG. 5.

[0086] 6 is a diagram showing an example of a product feature information storage unit of a user terminal. The product feature information stored in the product feature information storage unit 307 of this embodiment includes information items such as a store ID, a product ID, a product name, a type, and a unit price. The product feature information is information including values ​​for the items "store ID" and "product ID," as well as values ​​for other items. In this embodiment, the product feature information may be stored in advance in the product feature information storage unit 307 of the terminal device 300 by an administrator of the terminal device 300, etc.

[0087] In the product feature information, the value of the item "product ID" is identification information for identifying the product, the value of the item "product name" indicates the name of the product, the value of the item "type" indicates the type of product, and the value of the item "unit price" indicates the unit price of the product. Note that the information items included in the product feature information are not limited to the items shown in Figure 6, and may include items other than those shown in Figure 6, or may not include all of the items shown in Figure 6.

[0088] 7 is a diagram showing an example of the linked information storage unit of the user terminal. The log information (linked information) stored in the linked information storage unit 308 of this embodiment includes information items such as a user ID, date and time, a store ID, a product ID, and a quantity. The log information is information in which the user ID, the store ID, and the product ID are associated with each other.

[0089] In this embodiment, the log information may be stored in advance in the association information storage unit 308 of the terminal device 300 by an administrator of the terminal device 300 or the like.

[0090] In the log information, the value of the "Date and Time" field indicates the date and time when the user identified by the user ID purchased the product identified by the product ID at the store identified by the store ID. The value of the "Quantity" field indicates the number of products identified by the product ID purchased by the user identified by the user ID.

[0091] The information items included in the log information are not limited to the items shown in FIG. 7, and may include items other than those shown in FIG. 7, and may not include all of the items shown in FIG.

[0092] Furthermore, the log information in this embodiment may be POS data acquired from a point-of-sale information management system in each store identified by a store ID.

[0093] Furthermore, the user characteristic information, store characteristic information, product characteristic information, and log information of the present embodiment may be acquired from, for example, a server device that communicates with the terminal device 300, and may be updated periodically. The server device that communicates with the terminal device 300 may be a server device managed by each store or the like.

[0094] Next, the operation of the information processing system 100 of this embodiment will be described with reference to Fig. 8. Fig. 8 is a sequence diagram illustrating the operation of the information processing system. In the information processing system 100 of this embodiment, the terminal device 400 performs a predetermined anonymization process on the user characteristic information stored in the user characteristic information storage unit 405 of the terminal device 400 using the anonymization processing unit 410 (step S801). Next, the terminal device 400 transmits the anonymously processed information that has been subjected to the anonymization process to the information processing device 200 (step S802). The information processing device 200 acquires the anonymously processed information from the terminal device 400 using the information collection unit 250, and stores it in the anonymously processed information storage unit 210 (step S803).

[0095] Next, the information processing device 200 generates an estimation model using the model generation unit 260 (step S804). Details of the process of step S804 will be described later. Next, the model generation unit 260 of the information processing device 200 transmits the generated estimation model to the terminal device 300 (step S805).

[0096] When the terminal device 300 acquires the estimated model, the visualization processing unit 320 maps the multiple types of entities in which feature information is stored onto a vector space (step S806) and visualizes them (step S807). The details of the processing of steps S806 and S807 will be described later.

[0097] Next, the terminal device 300 creates a list using the list creation unit 330, distributes advertising information using the distribution processing unit 340, and predicts demand using the demand prediction unit 350 according to the user of the terminal device 300 (step S808).

[0098] In the example of FIG. 8, the processes from step S801 to step S808 are taken as an example of the process, but the process is not limited to this.

[0099] The processes from step S801 to step S803, the processes from step S804 to step S806, and the processes from step S807 and step S808 may be executed independently at different times.

[0100] Next, the processing of the model generation unit 260 of the information processing device 200 will be described with reference to Fig. 9. Fig. 9 is a flowchart shown by the information processing device. Fig. 9 shows details of step S804 in Fig. 8.

[0101] In the information processing device 200 of this embodiment, the model generation unit 260 acquires the anonymously processed information stored in the anonymously processed information storage unit 210 by the learning data generation unit 262 (step S901).

[0102] Next, the model generation unit 260 causes the learning data generation unit 262 to generate a data set (learning data) in which the anonymously processed information extracted in step S901 is associated with the user feature information stored in the user feature information storage unit 305 of the terminal device 300 (step S902). Note that the model generation unit 260 may acquire the user feature information stored in the user feature information storage unit 305 in advance and store it in the information processing device 200. Furthermore, for example, upon receiving an instruction to generate the estimation model 326, the model generation unit 260 may acquire the user feature information stored in the user feature information storage unit 305 from the terminal device 300. Furthermore, the user feature information stored in the user feature information storage unit 305 may be transmitted to the information processing device 200 together with the instruction to generate the estimation model.

[0103] Next, the model generation unit 260 constructs an estimation model by machine learning the data set generated in step S902 using the learning unit 263 (step S903). Next, the model generation unit 260 stores the constructed estimation model in the model storage unit 230 (step S904). Note that the estimation models stored in the model storage unit 230 are the same as the estimation models 326 held in the information extension unit 322 of the terminal device 300.

[0104] Here, the processing of the model generation unit 260 will be further explained with reference to Fig. 10. Fig. 10 is a diagram for explaining the processing of the model generation unit.

[0105] The anonymously processed information stored in the anonymously processed information storage unit 210 shown in Figure 10 is an example of anonymously processed information obtained by applying anonymization processing to the user characteristic information stored in the user characteristic information storage unit 405 of the terminal device 400.

[0106] Here, the learning data generation unit 262 sets the user characteristic information stored in the user characteristic information storage unit 305 of the terminal device 300, which is estimated to be similar to the anonymously processed information, and the anonymously processed information as a dataset.

[0107] In the example of Figure 10, the anonymously processed information including the user ID "a" in the anonymously processed information and the user characteristic information including the user ID "a" in the user characteristic information storage unit 305 are estimated to be similar, and therefore form a data set.

[0108] Also, in the example of Figure 10, among the user feature information stored in the user feature information storage unit 305, user feature information including user IDs "b" and "c" is considered to be user feature information that is estimated to be similar to the anonymously processed information, and becomes a data set together with the anonymously processed information.

[0109] The model generation unit 260 of this embodiment uses these data sets as training data and constructs an estimation model that learns the relationship between the anonymously processed information provided by the provider and the user characteristic information held by the user, using the learning unit 263.

[0110] Next, the processing of the terminal device 300 of this embodiment will be described with reference to Fig. 11 and Fig. 12. Fig. 11 is a first flowchart illustrating the processing of the terminal device. Fig. 11 shows details of the processing of the visualization processing unit 320 in step S807 of Fig. 8.

[0111] In the terminal device 300 of this embodiment, when the visualization processing unit 320 acquires the estimation model 326 from the information processing device 200, the information expansion unit 322 inputs the user feature information stored in the user feature information storage unit 305 to the estimation model 326 (step S1101), and acquires estimated user feature information output from the estimation model 326 (step S1102). In other words, the information expansion unit 322 is an example of an estimated information acquisition unit.

[0112] Specifically, the information expansion unit 322 inputs user characteristic information including the items "frequency," "time period," "alcoholic beverages," "beverages," and "lunch boxes" into the estimation model 326, and obtains estimated user characteristic information including the items "age," "gender," "eating out," "retail," and "manga."

[0113] Next, the visualization processing unit 320 uses the vector conversion unit 323 to map users, stores, and products into a vector space based on the user feature information stored in the user feature information storage unit 305, the store feature information stored in the store feature information storage unit 306, the product feature information stored in the product feature information storage unit 307, and the estimated user feature information acquired in step S1102 (step S1103). In this manner, in this embodiment, it is possible to acquire estimated user characteristic information estimated from anonymously processed information provided by a provider by using the estimation model 326. In other words, according to this embodiment, it is possible to acquire user characteristic information estimated based on the user characteristic information provided by the provider for a user whose user characteristic information the terminal device 300 holds.

[0114] In addition, in this embodiment, by using the estimation model 326, it is possible to expand the user characteristic information to be utilized while suppressing a significant loss in the number of populations and the influence of how permission is obtained.

[0115] 12 is a second flowchart illustrating the processing of the terminal device 320 in step S810 of FIG.

[0116] The terminal device 300 of this embodiment determines whether or not a display request for a vector space in which multiple types of entities are mapped has been received by the input receiving unit 321 of the visualization processing unit 320 (step S1201). If no display request has been received in step S1201, the terminal device 300 waits.

[0117] In step S1201, when a display request is received, the visualization processing unit 320 visualizes the vector space and displays it on the display device of the terminal device 300 (step S1202). In other words, the visualization processing unit 320 causes the screen generation unit 325 to generate screen data for displaying the vector space onto which multiple types of entities have been mapped by the vector conversion unit 323, and causes the terminal device 300 to display it.

[0118] Next, the visualization processing unit 320 determines whether or not an operation to select an entity has been performed on the terminal device 300 (step S1203). If an operation to select an entity has not been performed in step S1203, the visualization processing unit 320 may end the process.

[0119] When an operation to select an entity is performed in step S1203, the group generation unit 324 creates a group according to the selected entity (step S1204).

[0120] Specifically, the group generation unit 324 may group together entities that are different in type from the selected entity and have affinity with the selected entity. The affinity entities may be entities that are within a predetermined distance from the selected entity in the vector space.

[0121] Next, the visualization processing unit 320 causes the screen generating unit 325 to generate screen data for displaying information related to the generated group, and causes the terminal device 300 to display the screen data (step S1205).

[0122] In this way, according to this embodiment, the extended user feature information is stored in the entity relationship. Display examples of this embodiment will be described below with reference to Fig. 13 and Fig. 14. Fig. 13 is a first diagram showing a display example of a terminal device.

[0123] A screen 301 shown in FIG. 13 may be displayed on the terminal device 300 by the processing of step S1202 in FIG. 12, for example.

[0124] On screen 301, multiple types of entities, such as users, stores, and products, are each mapped onto a single vector space.

[0125] The screen 301 also includes a display area 91, in which information indicating the correspondence between the types of entities mapped onto the vector space and the dots displayed on the screen 301 is displayed.

[0126] Furthermore, the screen 301 of this embodiment displays an input field 92. In this embodiment, when identification information of an entity that serves as a search key is entered in the input field 92 and a search button is operated, information indicating the position of the entity whose identification information has been entered in the vector space displayed on the screen 301 may be displayed.

[0127] More specifically, for example, when a store ID is entered in the input field 92 as identification information for specifying an entity, the dot corresponding to the entered store ID may be displayed in a different manner from other dots on the screen 301. In this way, the user of the terminal device 300 can visually grasp the degree of affinity between the desired entity and other entities.

[0128] In addition, in this embodiment, when a dot is selected on the screen 301, feature information of the entity corresponding to the selected dot may be displayed on the screen 301.

[0129] In FIG. 13, display area 93 is an example of a display area that is displayed when dot D1 is selected, display area 94 is an example of a display area that is displayed when dot D2 is selected, and display area 95 is an example of a display area that is displayed when dot D3 is selected.

[0130] The type of entity indicated by the dot D1 is a user. The dot D1 also indicates a user identified by the user ID "a." Therefore, the display area 93 displays the user characteristic information associated with the user ID "a."

[0131] The type of entity indicated by dot D2 is a store. Dot D2 also indicates a store identified by store ID "3." Therefore, the display area 94 displays store characteristic information associated with store ID "3."

[0132] The type of entity indicated by the dot D3 is a product. The dot D3 also indicates a product identified by the product ID "11." Therefore, the display area 95 displays the product feature information associated with the product ID "11."

[0133] In this embodiment, when an entity is selected, the feature information of the selected entity is displayed. Therefore, according to this embodiment, the user of the terminal device 300 can understand the features of the selected entity.

[0134] In this embodiment, the screen 301 may include an operation button for instructing the generation of a group including a selected dot. In this case, when an operation for selecting a dot is performed on the screen 301 and then an operation for instructing the generation of a group is performed, the terminal device 300 may generate a group including an entity corresponding to the selected dot and entities that have a high affinity with this entity.

[0135] When a group is generated, the terminal device 300 may store, as group information, a list of identification information for identifying entities included in the group.

[0136] In this way, by generating a group including the selected entity, it is possible to allow the user of the terminal device 300 to know, for example, stores, products, etc. that have a high affinity with a certain user.

[0137] In this embodiment, for example, when an instruction to create a group is received, the types of entities included in the group may be identified, which makes it possible to identify users whose attributes are similar to those of a certain user.

[0138] Furthermore, in this embodiment, when entities are grouped, the terminal device 300 may display the results of statistical processing performed on the feature information of each of the grouped entities as information about the group.

[0139] 14 is a second diagram showing an example of a display on the terminal device 300. The screen 302 shown in FIG. 14 may be displayed on the terminal device 300 by, for example, the processing in step S1205 of FIG.

[0140] The screen 302 includes display areas 101 and 102. The display area 101 may display a list of generated groups. The display area 102 may display information about a group selected from the list displayed in the display area 101.

[0141] In the example of FIG. 14, "Group AB" is selected from the list of groups displayed in display area 101, and information about group AB is displayed in display area 102.

[0142] Specifically, the information about the group may be, for example, information indicating the number of users who have purchased a certain product, the total purchase amount of the product for each user, etc. By displaying such information, the user of the terminal device 300 can understand the affinity between the product characteristics indicated by the product characteristic information and the user characteristics indicated by the user characteristic information. Therefore, it is possible to provide the user of the terminal device 300 with information about suppliers, etc., and to propose campaigns, etc.

[0143] Furthermore, the information about the group may include the number of visitors by age group at a certain store. By displaying such information, the user of the terminal device 300 can understand the affinity between the store characteristics indicated by the store characteristic information and the user characteristics indicated by the user characteristic information. Therefore, in this embodiment, the user of the terminal device 300 can be assisted in appropriately using outdoor advertisements near the store and advertisements at the store entrance and inside the store.

[0144] In this embodiment, the entity types are users, stores, and products, but the entity types in this embodiment are not limited to these. The entity types in this embodiment may be any type as long as linking information linking the feature information of each entity can be obtained.

[0145] 13 and 14 are displayed on the terminal device 300 in this embodiment, but the present invention is not limited to this. The screens 301 and 302 may be displayed on other terminal devices managed by the business operator that manages the terminal device 300.

[0146] As described above, according to this embodiment, if the user of the terminal device 300 is a store manager or the like, he or she can acquire hypotheses for improving stocking amounts and shelf allocations.

[0147] Next, a description will be given of the processing of the list generation unit 330 of this embodiment. For example, the list generation unit 330 of this embodiment may generate a list in which users included in a group by the group generation unit 324 are the target users of advertising information.

[0148] In this embodiment, by doing so, it is possible to create a user group using either or both of the user's user characteristic information and the estimated user characteristic information, and to list the users included in the target group. This makes it possible to create a list under more flexible conditions that also include the provider's user characteristic information.

[0149] Next, a description will be given of the distribution processing unit 340. The distribution processing unit 340 of this embodiment can distribute advertising information of specific content to specific users based on the list created by the list generating unit 330.

[0150] For example, the distribution processing unit 340 may use machine learning to learn the correspondence between the user characteristics indicated by the user characteristic information and estimated user characteristic information held by the user (the business operator managing the terminal device 300) and an effect index indicating an index of the effect obtained by advertising information previously distributed to the user, and generate a distribution model. The effect index may be, for example, a click rate or a purchase rate. For example, this distribution model may output advertising information to be distributed to a user to whom first user characteristic information is input and whose characteristics are indicated by the first user characteristic information.

[0151] In this embodiment, by using estimated user characteristic information in this manner, it is possible to learn perspectives that could not be obtained from the user characteristic information held by the user alone, thereby enabling effective information delivery.

[0152] Next, a description will be given of the demand forecasting unit 350. The demand forecasting unit 350 of this embodiment forecasts the amount of a certain product to be purchased in a certain period of time.

[0153] For example, the demand forecasting unit 350 may use user characteristic information held by the user and estimated user characteristic information to learn correspondences between user behavior, products purchased by the user, the quantity of products purchased, location, season, etc. through machine learning, and generate a demand forecasting model.

[0154] In this embodiment, by doing so, it is possible to build a highly accurate demand forecasting model that has a deeper understanding of the user and can respond to changes in circumstances, locations, seasons, etc.

[0155] Furthermore, by building such a demand forecasting model, it becomes possible to adjust order quantities, reduce inventory loss and stockout risks, and optimize inventory distribution.

[0156] Although the present invention has been described above based on the embodiments, the present invention is not limited to the requirements shown in the above embodiments. These requirements can be changed without departing from the spirit of the present invention, and can be appropriately determined depending on the application form. [Explanation of symbols]

[0157] 100 Information Processing Systems 200 Information processing device 210 Anonymously processed information storage unit 230 Model Memory Unit 250 Information Gathering Department 260 Model Generation Unit 300, 400 terminal equipment 305, 405 User characteristic information storage unit 306, 406 Store characteristic information storage unit 307, 407 Product feature information storage unit 308, 408 Linking information storage unit 320 Visualization Processing Unit 330 List Generation Unit 340 Distribution Processing Unit 350 Demand Forecasting Department 410 Anonymous Processing Unit

Claims

1. An information processing system including an information processing device, a first terminal device that stores first user characteristic information, and a second terminal device that stores second user characteristic information, The information processing device includes: an information collection unit that collects, from the second terminal device, anonymously processed information obtained by performing a predetermined anonymization process on the second user characteristic information; a learning unit that uses the first user feature information and the anonymously processed information as learning data to perform machine learning on a relationship between the first user feature information and the anonymously processed information, and that, when the first user feature information is input, generates a model that outputs estimated user feature information that is estimated from the relationship between the first user feature information and the anonymously processed information; a model output unit that outputs the model to the first terminal device.

2. The predetermined anonymization processing is a process for anonymizing the second user characteristic information so as to satisfy requirements stipulated in the Personal Information Protection Act; The information processing system according to claim 1 , further comprising a process of anonymizing the second user characteristic information so as to satisfy k-anonymity.

3. The first terminal device an estimated information acquisition unit that inputs the first user feature information to the model and acquires the estimated user feature information; a vector conversion unit that refers to a storage unit that stores feature information indicating features of each of a plurality of types of objects, including the first user feature information and the estimated user feature information, and linking information that links the plurality of types of objects, and converts the plurality of types of objects into vector representations and maps them into a single vector space; 3. The information processing system according to claim 1, further comprising: a display control unit that causes a display device to display a screen including the vector space onto which the plurality of types of objects are mapped.

4. The first terminal device The information processing system according to claim 3, further comprising a distribution processing unit that generates a machine-learned model of the relationship between the user and the advertising information using the first user characteristic information, the estimated user characteristic information, and an effectiveness indicator that indicates an indicator of the effectiveness obtained by the advertising information distributed to the user as learning data.

5. The first terminal device The information processing system according to claim 3, further comprising a demand forecasting unit that uses machine learning to learn correspondences between user behavior, products purchased by the user, and quantities of products purchased, which are contained in the first user characteristic information and the estimated user characteristic information, and generates a demand forecasting model.

6. An information processing device that communicates with a first terminal device that holds first user characteristic information and a second terminal device that holds second user characteristic information, an information collection unit that collects, from the second terminal device, anonymously processed information obtained by performing a predetermined anonymization process on the second user characteristic information; a learning unit that uses the first user feature information and the anonymously processed information as learning data to perform machine learning on a relationship between the first user feature information and the anonymously processed information, and that, when the first user feature information is input, generates a model that outputs estimated user feature information that is estimated from the relationship between the first user feature information and the anonymously processed information; a model output unit that outputs the model to the first terminal device.

7. An information processing method by an information processing device that communicates with a first terminal device that stores first user characteristic information and a second terminal device that stores second user characteristic information, The information processing device, collecting, from the second terminal device, anonymously processed information obtained by performing a predetermined anonymization process on the second user characteristic information; using the first user feature information and the anonymously processed information as learning data, machine learning is performed on a relationship between the first user feature information and the anonymously processed information, and when the first user feature information is input, a model is generated that outputs estimated user feature information estimated from the relationship between the first user feature information and the anonymously processed information; and outputting the model to the first terminal device.

8. An information processing device that communicates with a first terminal device that holds first user characteristic information and a second terminal device that holds second user characteristic information, collecting, from the second terminal device, anonymously processed information obtained by performing a predetermined anonymization process on the second user characteristic information; using the first user feature information and the anonymously processed information as learning data, machine learning is performed on a relationship between the first user feature information and the anonymously processed information, and when the first user feature information is input, a model is generated that outputs estimated user feature information estimated from the relationship between the first user feature information and the anonymously processed information; A program that outputs the model to the first terminal device and executes processing.

Images