JP2023001109A - Information processing device, information processing method, and information processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a mechanism in which a data zone affiliated enterprise wants to provide its own data.

SOLUTION: An information processing device comprises: an acquisition unit which acquires anonymous information obtained by vectorizing customer data from each of a plurality of enterprises; a learning unit which performs machine learning of AI using anonymous information from each enterprise; an estimation unit which estimates a customer suited to the objective of a specific enterprise among a plurality of enterprises by the AI which has experienced the machine learning; and a provision unit which provides information based on a result of the estimation made by the estimation unit.

SELECTED DRAWING: Figure 1

COPYRIGHT: (C)2023,JPO&INPIT

Description

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

In a data sharing system, a technology is disclosed for safely using the data held by each company without disclosing it to other companies.

Japanese Patent No. 6803598

However, the above-described conventional technology uses a predetermined cryptographic method (a cryptographic method that allows operations including search and analysis to be performed in an encrypted state) for sensitive data provided by organizations participating in a data sharing system. However, the key used for encryption and decryption is simply managed by each organization that provided the sensitive data.

In addition, there is a problem that if all clients are forced to integrate IDs (ID Connect) using a common ID, the number of companies participating in the data area will not increase. Also, even if companies join the data sphere, it would be unrealistic to have each company send all customer data (personal information) to the mother server side (many companies would not agree with this). Therefore, it is necessary to design so that the member companies of the data zone do not release much of their own data and cannot rely only on the data of other companies for marketing purposes. In other words, it is necessary to have a mechanism that prevents free riding and a mechanism that makes people want to release their company's data.

The present application has been made in view of the above, and aims to provide a mechanism that makes data area member companies want to publish their own data.

An information processing apparatus according to the present application includes an acquisition unit that acquires anonymous information obtained by vectorizing customer data from each of a plurality of companies, a learning unit that performs AI machine learning using the anonymous information from each company, and An estimating unit for estimating a customer suitable for the purpose of a specific company among the plurality of companies by AI that has undergone machine learning, and a providing unit for providing information based on the estimation result of the estimating unit. .

According to one aspect of the embodiments, it is possible to provide a mechanism that makes data zone member companies want to publish their own data.

FIG. 1 is an explanatory diagram showing an outline of an information processing method according to an embodiment. FIG. 2 is an explanatory diagram showing an outline of AI learning on the server side. FIG. 3 is an explanatory diagram showing an example of learning from the campaign purpose. FIG. 4 is an explanatory diagram showing an outline of vectorization of customer data. FIG. 5 is an explanatory diagram showing an outline of inference on the server side. FIG. 6 is a schematic diagram that simplifies inference on the server side. FIG. 7 is an explanatory diagram showing an outline of vector neighborhood customer extension. FIG. 8 is a conceptual diagram showing an image of the 1stParty data zone concept. FIG. 9 is a diagram illustrating a configuration example of a mother server according to the embodiment; FIG. 10 is a diagram showing an example of a user information database. FIG. 11 is a diagram showing an example of the history information database. FIG. 12 is a diagram showing an example of an anonymous information database. FIG. 13 is a flowchart illustrating a processing procedure according to the embodiment; FIG. 14 is a diagram illustrating an example of a hardware configuration;

Embodiments for implementing an information processing apparatus, an information processing method, and an information processing program according to the present application (hereinafter referred to as "embodiments") will be described in detail below with reference to the drawings. The information processing apparatus, information processing method, and information processing program according to the present application are not limited to this embodiment. Also, in the following embodiments, the same parts are denoted by the same reference numerals, and overlapping descriptions are omitted.

[1. Outline of information processing method]
First, an outline of an information processing method performed by an information processing apparatus according to an embodiment will be described with reference to FIG. FIG. 1 is an explanatory diagram showing an outline of an information processing method according to an embodiment. In FIG. 1, a case where data zone member companies provide a system that makes them want to release their own data will be explained as an example.

As shown in FIG. 1, an information processing system 1 according to the embodiment includes a client terminal 10 and a mother server 100. The client terminal 10 and the mother server 100 are connected to each other via a network N by wire or wirelessly so that they can communicate with each other. The network N is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network) such as the Internet.

In addition, the client terminal 10 can be a wireless communication network such as LTE (Long Term Evolution), 4G (4th Generation), 5G (5th Generation: fifth generation mobile communication system), Bluetooth (registered trademark), wireless LAN (Local It is possible to connect to the network N via short-range wireless communication such as Area Network) and communicate with the mother server 100 .

The client terminal 10 is an information processing device used by each company on the client side (data zone member company). For example, the client terminal 10 is a PC (Personal Computer), a server device, or a smart device such as a smart phone or a tablet terminal. It should be noted that each company on the client side is preferably a company in a different industry. Of course, competing companies (companies in the same industry) may be included in the data area member companies.

The client terminal 10 of each company collects customer data on each user U (user) who is a customer of the company. For example, the client terminal 10 of each company collects, as customer data, registration information, store visit records, product purchase history/service usage history, questionnaire responses, etc. of each user U who purchased products/services of the company.

In the example shown in FIG. 1, for identification, the client terminal 10 used by company A is "client terminal 10A", the client terminal 10 used by company B is "client terminal 10B", and the client terminal 10 used by company C is "client terminal 10A". The terminal 10 is called a "client terminal 10C".

The mother server 100 is, for example, a PC, a server device, a mainframe, a workstation, or the like. Note that the mother server 100 may be realized by cloud computing. Also, the mother server 100 may collect customer data (mother-side data) on each user U as a customer who uses the mother-side service, separately from the client terminals 10 of each company.

For example, the mother server 100 cooperates with the terminal device of each user U, and provides API (Application Programming Interface) services for various applications (hereinafter referred to as apps), etc., and various data may be provided.

Also, the mother server 100 may be an information processing device that provides some kind of web service online to each user U's terminal device. For example, the mother server 100 provides web services such as Internet connection, search service, SNS (Social Networking Service), electronic commerce, electronic payment, online games, online banking, online trading, accommodation/ticket reservation, video/music distribution, news, etc. , maps, route search, route guidance, route information, operation information, weather forecast, and other services. In practice, the mother server 100 may cooperate with various servers that provide web services as described above, mediate the web services, or take charge of the processing of the web services.

Note that the mother server 100 can acquire user information about the user U. FIG. For example, the mother server 100 acquires information about attributes of the user U, such as the user U's sex, age, and area of residence. The mother server 100 stores and manages identification information indicating the user U (user ID, etc.) and information about the attributes of the user U. FIG.

Also, the mother server 100 acquires various types of history information (log data) indicating the actions of the user U from the terminal device of the user U or from various servers based on the user ID or the like. For example, the mother server 100 acquires a location history, which is a history of the location and date and time of the user U, from the terminal device. The mother server 100 also acquires a search history, which is a history of search queries input by the user U, from a search server (search engine). Also, the mother server 100 acquires a viewing history, which is a history of content viewed by the user U, from the content server. The mother server 100 also acquires the purchase history (payment history), which is the history of user U's product purchases and payment processing, from the electronic commerce server and the payment processing server. Further, the mother server 100 may acquire the exhibition history and sales history, which are the history of the user U's exhibition in the marketplace, from the electronic commerce server or the payment server. In addition, the mother server 100 acquires the posting history, which is the posting history of the user U, from a posting server that provides a word-of-mouth posting service or an SNS server.

Note that the number of devices included in the information processing system 1 shown in FIG. 1 is not limited to that illustrated. For example, in FIG. 1, only three client terminals 10 are shown for simplification of illustration, but this is only an example and is not limited, and may be four or more. Also, the mother server 100 may be distributed among a plurality of server devices.

For example, as shown in FIG. 1, the mother server 100 pre-connects the client terminals 10 (10A, 10B, 10C) of each company participating in the data area via the network N in advance (when each company joins the data area, etc.). , . . . ) for anonymizing (vectorizing) customer data (step S1).

At this time, the mother server 100 may customize AI for each company. That is, the mother server 100 may distribute different AIs to each company. Since the content of information acquired as customer data may differ depending on the type of business and business type of the company, it is possible to change the specifications of the AI according to the content of the customer data acquired, and it is not possible to use the same AI uniformly. may It is sufficient if the mother server 100 (server side) comprehends and manages the AI specifications distributed to each company.

Subsequently, the mother server 100 collects anonymous information obtained by anonymizing (vectorizing) the customer data of each company by AI from the client terminals 10 of each company participating in the data zone (step S2).

In the present embodiment, the client terminal 10 of each company acquires customer data relating to the company's customers, and generates anonymous information by converting (vectoring) the acquired customer data into a vector using AI. At this time, the client terminal 10 may perform so-called embedding to generate an embedding vector based on customer data as anonymous information. Then, the client terminal 10 of each company transmits anonymous information obtained by anonymizing (vectorizing) customer data to the mother server 100 in response to a request from the mother server 100 or voluntarily.

For example, the client terminal 10 uses AI to input customer data into a vector conversion model constructed using a machine learning technique using a neural network on the mother server 100 (server side), and transforms the customer data into a multidimensional model. Convert to a real-valued vector. Alternatively, the client terminal 10 may convert customer data into a vector according to a predetermined conversion rule (rule base).

The anonymous information is irreversible information, and the original customer data cannot be restored from the vector value of the anonymous information. Here, "cannot be restored" means not only the concept that the original data cannot be deterministically calculated from the vector value, but also the original data such as those that cannot be estimated, that is, the customer data It also includes the concept that it is difficult to calculate or estimate what it was like. That is, anonymous information cannot be decrypted like encrypted data. For example, when customer data is anonymized (vectorized), it is possible to place it in predetermined groups or segments based on common attributes, etc., or to "round" numbers by performing rounding (rounding off, rounding down, etc.). . Moreover, customer data and vector values are not limited to a one-to-one relationship in which the same vector value is obtained from the same data, but a many-to-one relationship in which the same vector value is obtained from different data. There is also a many-to-many relationship in which different vector values are obtained depending on the combination of data even if the same data is included. Furthermore, when different AIs are distributed to different companies, the meaning of the vector value itself may differ depending on the company. Therefore, one vector value is not necessarily based on the same data. Also, what the mother server 100 (server side) needs is the vector value, not the customer data that is the basis (source) of the vector value.

In addition, since it is the mother server 100 (server side) that distributes AI that defines the vector value of anonymous information and the content that it means, even if each company or third party sees only the vector value of anonymous information, , it is not possible to understand what the vector values mean, so even if anonymous information leaks out, customer data will not be leaked, so it is safe. Note that each time a vector is created, even if it is for the same user, it becomes a different vector, and the vector is not necessarily the same each time.

Subsequently, the mother server 100 uses all the anonymous information collected from the client terminals 10 of each company to learn the purpose setting data as correct answers (step S3). For example, the mother server 100 performs machine learning using anonymous information of all customers of all companies participating in the data zone to generate a customer evaluation model. At this time, the mother server 100 may learn using customer data (mother-side data) independently acquired by the mother side in addition to all anonymous information. Note that the mother server 100 may generate anonymous information by anonymizing (vectorizing) the mother-side data using AI.

Subsequently, the mother server 100 makes an inference for each company from the learning result (step S4). For example, the mother server 100 generates a list of customers who are likely to purchase a specific product from a specific company based on the results of learning using all anonymous information collected from the client terminals 10 of each company. At this time, the mother server 100 may input anonymous information from a specific company into the customer evaluation model, and obtain a list of customers who are likely to purchase a specific product of the specific company as an output.

Subsequently, the mother server 100 provides the client terminal 10 of each company with the inference result corresponding to each company and information based on the inference result via the network N (step S5). For example, the mother server 100 provides the client terminal 10 of a specific company with a list of customers who are likely to purchase a specific product of a specific company.

[1-1. AI learning on the server side]
Next, AI learning in the mother server 100 (server side) according to this embodiment will be described with reference to FIG. FIG. 2 is an explanatory diagram showing an outline of AI learning on the server side.

As shown in FIG. 2, the data area consists of a CDP (Customer Data Platform) on the client side and a server on the mother side. That is, the configuration consists of the client terminal 10 of each company and the mother server 100 .

In the CDP on the client side, the client terminal 10 of each company generates anonymous information by converting (vectoring) customer data (1st party data) owned by the company into vectors using AI. At this time, the client terminal 10 of each company generates a vector for each ID of the company as anonymous information. That is, a combination of the customer ID of each company and the vector data is generated.

The customer data held by each company differs from company to company. For example, Company A collects information such as "visit", "purchase", and "requests/conversations during business negotiations" as customer data. Company B collects information such as "product purchase", "questionnaire response", and "web advertisement exposure" as customer data. Company C collects information such as "search", "news contact", and "purchased products" as customer data. However, these are only examples.

Then, the client terminal 10 of each company transmits the generated anonymous information as transmission data to the mother server 100 (server side) in response to a request from the mother server 100 or spontaneously. For security reasons, the client terminal 10 of each company may further encrypt the generated anonymous information and transmit it to the mother server 100 (server side). Encrypting and transmitting data itself is possible within the scope of common technical knowledge. Alternatively, tunneling may be used to establish a closed virtual direct line between the mother side and the client side. For example, a VPN (Virtual Private Network) may be constructed between the mother side and the client side.

At this time, the client terminal 10 of each company transmits only the vector data associated with the common ID as anonymous information. Common ID is a common customer ID in the data area. The vector data associated with the common ID is the vector data of the customer who has given permission to provide data to the data area. As a whole, cooperation between the mother side and the client side is exchange of vector data only for users who have permitted common ID cooperation. It should be noted that only vector data (anonymous information) is transmitted, and customer data itself is not transmitted. Even if it is vectorized, it may be interpreted as personal information, but it is overwhelmingly safer than raw personal information data. Although it is known from the common ID that they are the same person, it is impossible to distinguish them because information about individuals such as place of residence, age, and sex is not included or is anonymized (vectorized).

Moreover, it is not necessary to associate the vector data of all customers with the common ID. For example, if a company only participates in the data zone without obtaining individual customers' permission to use data within the data zone, there is no need to associate the common ID with the customer's vector data.

In this embodiment, the ID integration (ID Connect) with the common ID is set to "arbitrary". In other words, even if there is a customer who "does not" integrate IDs (ID Connect) with a common ID, the design should function as a whole.

In addition, it will be designed so that the purpose can be achieved by not sending raw customer data (personal information) to the mother server from companies that have joined the data zone. That is, raw customer data (personal information) owned by each company is not sent to the mother server 100 . Raw customer data (personal information) owned by each company is anonymized (vectorized) and sent to the mother server 100 together with an ID.

In addition, by introducing decentralized AI that anonymizes (vectorizes) the company's customer data using only the company's own data, the company's targeting accuracy does not improve unless the company's data is produced in both quality and quantity. and This prevents free rides that use only other companies' data without releasing their own data.

It should be noted that the greater the quantity and quality of the company's customer data, the greater the benefits it receives. In other words, the more vector data linked with the common ID, the higher the accuracy, as the use permission for data within the data area is obtained on a customer-by-customer basis.

Each customer consents to "anonymization (vectorization) of customer data concerning itself and use within the data area". In addition, each company may give an incentive or privilege to each customer's approval. For example, if each customer approves, a discount coupon or the like is sent from each company within the data area. In addition, licensed customers will be issued rewards according to their activities within the data zone (purchasing products/services of companies participating in the data zone, contacting advertisements, answering questionnaires, etc.).

The mother server 100 (server side) implements AI learning by setting a list of past product purchaser IDs from each company's CDP as correct data. For example, when company A wants to create a product purchase probability list (recommendation list), which is a list of customers who are likely to purchase product X of its own company, the mother server 100 (server side) obtains past product X from company A's CDP. AI learning is performed by setting the purchaser's ID list as correct data as the purpose.

Now, with reference to FIG. 3, an example of learning from campaign objectives will be described. FIG. 3 is an explanatory diagram showing an example of learning from the campaign purpose. In addition, AI learning shall be implemented for every campaign. As shown in FIG. 3, the mother server 100 (server side) implements AI learning by setting a list of IDs of past purchasers of product X from company A's data.

At this time, the mother server 100 (server side) may set customer data (mother side data) owned by itself separately from each company's anonymous information in consideration of the correct data. For example, if company A is an automobile manufacturer and product X is a specific vehicle, the mother server 100 (server side) adds customer data of "persons interested in golf" to correct data as mother side data. can be set as

Then, the mother server 100 (server side) provides all records of the customer data of the company to which the correct data is given, and all records of other member companies linked by the common ID among all records of the customer data of the company. is used as learning data, AI learning is performed.

[1-2. Vectorization of customer data]
Next, vectorization of customer data according to this embodiment will be described with reference to FIG. FIG. 4 is an explanatory diagram showing an outline of vectorization of customer data. In this embodiment, the personal information of each customer of each company is vectorized so that it is not personal information, but the customer characteristics can be processed.

As shown in FIG. 4, for example, when company A is an automobile manufacturer and product X is a specific vehicle, the client terminal 10 of company A stores, as customer data, purchased products, store visit/conversation content, inquiry/catalog request , Owned WEB logs, contact with advertisements, contact with sales promotions, etc.

In addition, the client terminal 10 of Company A uses one of the following methods (1) and (2) as a method of vectorizing customer data.

(1) The client terminal 10 of Company A assumes that the products (models of cars in the case of cars) that the customer negotiates with or purchases at the same time are products with similar purposes, and vectorizes the products. Next, using the product vector, two customers are randomly extracted, and learning is performed using the product vector near and far as a positive example and a negative example.

(2) The client terminal 10 of Company A learns from a fill-in-the-blank problem using a BERT (Bidirectional Encoder Representations from Transformers) model and acquires from the embedding layer. Note that BERT is a Transformer-based machine learning technique for pre-training of Natural Language Processing (NLP). Actually, it is not limited to BERT, and other natural language processing may be used. For example, it may be a distributed representation (word vector) such as Word2Vec. Distributed representation is to embed characters/words in a vector space and treat them as one point in the space, and is also called word embedding.

FIG. 4 shows an example of anonymization (vectorization) of company A's customer data. Here, for simplification of explanation, the dimension of the vector is expressed in two dimensions, and the vector group is classified into several segments according to the nearness and farness of the vector. For example, they are classified into segments such as "luxury sports car oriented", "luxury SUV oriented", and "low price range oriented".

In the example of the anonymization (vectorization) of customer data of company A shown in FIG. Each customer is expressed in a multi-dimensional manner with all elements taken into consideration.

[1-3. Inference on the server side]
Next, inference on the server side according to this embodiment will be described with reference to FIGS. 5 and 6. FIG. FIG. 5 is an explanatory diagram showing an outline of inference on the server side. FIG. 6 is a schematic diagram that simplifies inference on the server side. Inference on the server side may be performed after or at the same time as AI learning on the server side.

As shown in FIG. 5, the mother server 100 (server side) inputs the customer data vector (anonymous information) of each company to AI on the mother side, and creates a product purchase probability list (recommendation list) for each customer of each company. ).

At this time, the mother server 100 (server side) provides a purchase probability list of a specific product by ID of a specific company (for example, product X purchase probability A company ID list) for only customers of a specific company, and a list for customers of all companies. A list of purchase probabilities of specific products (for example, a list of common IDs for product X purchase probabilities) based on common IDs such as product X is obtained.

Then, the mother server 100 (server side) transmits the purchase probability list of the specific product by the ID of the specific company targeted only for the customers of the specific company only to the client terminal 10 of the specific company. At this time, the client terminal 10 of the specific company performs processing of vector neighborhood customer expansion, which will be described later, based on the purchase probability list of the specific product by the ID of the specific company targeted only for the customers of the specific company.

In addition, the mother server 100 (server side) distributes advertisements to customers of all companies, distributes coupons on SNS, etc., based on the purchase probability list of a specific product with a common ID for customers of all companies. can do

As shown in FIG. 6, when Company A desires a list of customers with a high probability of purchasing Company A's product X, the mother server 100 (on the server side) has purchased Company A's product X in the past. Among the customers who have accepted the common ID, the information of the customers who have approved the shared ID cooperation is input to AI on the mother side as correct data, and a purchase probability list (recommendation list) of product X of company A is acquired. Then, the mother server 100 (server side) carries out advertisement distribution, coupon distribution on SNS, etc., based on the purchase probability list of product X. FIG. In addition, the mother server 100 (server side) may return the list to Company A's CRM (Customer Relationship Management) and perform vector neighborhood expansion in Company A's CDP or the like. Allows you to conduct marketing activities in your company's CRM.

[1-4. Vector Neighborhood Customer Extension]
Next, vector neighborhood customer expansion according to the present embodiment will be described with reference to FIG. FIG. 7 is an explanatory diagram showing an outline of vector neighborhood customer extension.

When inferring the correct data on the mother server 100 (server side), only vector data on the mother side (vector data of customers who are ID Connected) is used as the basis, so when returning to the client side CDP, Perform neighborhood expansion on vectors of in-house data.

For example, the client terminal 10 of company A assumes that users U1 and U2 who are customers are candidates for purchasing product X, based on company A's customer data. Then, the client terminal 10 of Company A executes vectorization learning processing within the CDP of Company A (see "Customer Vector of Company A" in FIG. 7).

At this time, the mother server 100 (server side) utilizes the data of the users U1 and U4, among the users U1 to U4, who are performing ID integration (ID Connect) with the common ID, and uses the AI on the mother side to learn. Here, users U1 to U4 are explained as an example, but in reality, the number of data including other companies' customers (ID Connected users) is much larger.

The mother server 100 (server side) infers a person who has a high probability of purchasing product X of company A from the vectors of all customers of all companies on the mother side. At this time, user U1 is extracted from the data of company A (user U2 is not ID-connected).

Then, the mother server 100 (server side) feeds back U1 to the CDP of company A as a product X purchase candidate. Therefore, the client terminal 10 of company A performs vector neighborhood customer expansion according to the feedback from the mother side. Specifically, the client terminal 10 of company A has user U2 in the vicinity of user U1 in the customer data vector of company A, as shown in "A company customer vector" in FIG. , user U2 is also a purchase candidate for product X.

[1-5. 1stParty data zone concept]
As mentioned above, when each company joins the 1stParty data zone concept, extremely high-precision personal marketing becomes possible. FIG. 8 is a conceptual diagram showing an image of the 1stParty data zone concept.

For example, targeting can be done with AI that has learned all customer data that has accepted the common ID Connect of all companies that join the data zone (high-precision targeting).

In addition, the customer data of each company will not be passed on to other companies or the mother side. The customer data of each company is converted into a list of numbers (vectorized), and in the state of anonymous information, AI is learned on the mother side (the company's customer data itself is OK).

Also, the higher the quantity and quality of the company's vector data (anonymous information based on the company's customer data), the higher the company's targeting accuracy. Therefore, the more the company's vector data provided to the 1stParty data zone, the higher the accuracy. Conversely, the smaller the number of company vector data, the lower the accuracy (a mechanism that cannot be free-rided).

If you do not provide your company's vector data at all, there will only be vector data of other companies whose content is unknown in the data area, and the relationship between your company's customer data and your company's vector data is unknown. By providing your company's vector data, you can target by linking your company's vector data with the vector data of other companies. In addition, the more the content and number of the company's vector data increases, the more connections there are with the vector data of other companies, and the more information can be obtained from the data area.

In order to improve their own targeting accuracy, each company actively provides its own vector data, creating a virtuous cycle in which a large amount of vector data naturally gathers on the mother side.

According to this embodiment, data area cooperation and targeting accuracy can be improved without each company providing raw personal information. In addition, since there is a mechanism that prevents free riding, the accuracy as a whole improves synergistically. In addition, when viewed from the perspective of each company, even if it is in the data zone, other companies cannot see their company's data and situation (for example, from the perspective of company A, how many customers have visited their own store, etc. unknown to the company). In addition, if the mother side is developing a large number of services (portal site, etc.), it can also be used for mass marketing such as advertisement distribution and SNS.

[2. Mother server configuration example]
Next, the configuration of the mother server 100 according to the embodiment will be described using FIG. FIG. 9 is a diagram showing a configuration example of the mother server 100 according to the embodiment. As shown in FIG. 9, the mother server 100 has a communication section 110, a storage section 120, and a control section .

(Communication unit 110)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card) or the like. Also, the communication unit 110 is connected to the network N by wire or wirelessly.

(storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or flash memory, or a storage device such as a hard disk or an optical disk. As shown in FIG. 9 , storage unit 120 has user information database 121 , history information database 122 , and anonymous information database 123 .

(User information database 121)
The user information database 121 stores user information about users U (users) that the mother server 100 independently collects. For example, the user information database 121 stores various information such as user U attributes. FIG. 10 is a diagram showing an example of the user information database 121. As shown in FIG. In the example shown in FIG. 10, the user information database 121 has items such as "user ID (Identifier)", "age", "gender", "home", "place of work", and "interest".

“User ID” indicates identification information for identifying the user U. The “user ID” may be the user U's contact information (telephone number, e-mail address, etc.), or may be identification information for identifying the user U's client terminal 10 .

"Age" indicates the age of the user U identified by the user ID. Note that the "age" may be information indicating a specific age of the user U (for example, 35 years old) or information indicating the age of the user U (for example, 30's). . Alternatively, the "age" may be information indicating the date of birth of the user U, or information indicating the generation of the user U (for example, born in the 80's). "Gender" indicates the gender of the user U identified by the user ID.

"Home" indicates location information of the home of the user U identified by the user ID. In the example shown in FIG. 10, "home" is represented by an abstract code such as "LC11", but may be latitude/longitude information or the like. Also, for example, "home" may be an area name or an address.

"Place of work" indicates location information of the place of work (school in the case of a student) of the user U identified by the user ID. In the example shown in FIG. 10, the "place of work" is illustrated as an abstract code such as "LC12", but may be latitude/longitude information or the like. Also, for example, the "place of work" may be an area name or an address.

"Interest" indicates the interest of the user U identified by the user ID. That is, "interest" indicates an object in which the user U identified by the user ID is highly interested. For example, the "interest" may be a search query (keyword) that the user U has entered into a search engine and searched for. In the example shown in FIG. 10, one "interest" is shown for each user U, but there may be more than one.

For example, in the example shown in FIG. 10, the age of the user U identified by the user ID "U1" is "twenties" and the gender is "male". Also, for example, the user U identified by the user ID "U1" indicates that the home is "LC11". Also, for example, the user U identified by the user ID "U1" indicates that the place of work is "LC12". Also, for example, the user U identified by the user ID "U1" indicates that he is interested in "sports".

Here, in the example shown in FIG. 10, abstract values such as “U1”, “LC11” and “LC12” are used, but “U1”, “LC11” and “LC12” are concrete values. It is assumed that information such as character strings and numerical values is stored. Hereinafter, abstract values may also be illustrated in diagrams relating to other information.

The user information database 121 is not limited to the above, and may store various types of information depending on the purpose. For example, the user information database 121 may store various information about the client terminal 10 of the user U. FIG. In addition, the user information database 121 stores user U's demographics (demographic attributes), psychographics (psychological attributes), geographics (geographical attributes), behavioral attributes (behavioral attributes), etc. Information about attributes may be stored. For example, the user information database 121 includes name, family structure, hometown (local), occupation, position, income, qualification, residence type (detached house, condominium, etc.), presence or absence of car, commuting time, commuting time, commuting time. Information such as routes, commuter pass sections (stations, lines, etc.), frequently used stations (other than the nearest station to your home or place of work), lessons (places, time zones, etc.), hobbies, interests, lifestyle, etc. may

(History information database 122)
The history information database 122 stores various types of information related to history information (log data) indicating the behavior of the user U, which the mother server 100 independently collects. FIG. 11 is a diagram showing an example of the history information database 122. As shown in FIG. In the example shown in FIG. 11, the history information database 122 has items such as "user ID", "location history", "search history", "browsing history", "purchase history", and "posting history".

“User ID” indicates identification information for identifying the user U. "Position history" indicates a position history, which is a history of the user's U position and movement. Also, "search history" indicates a search history that is a history of search queries input by the user U. FIG. "Browsing history" indicates a browsing history that is a history of contents browsed by the user U. FIG. "Purchase history" indicates the purchase history of the user U's purchases. In addition, “posting history” indicates a posting history that is a history of posts by the user U. FIG. In addition, the “posting history” may include questions about user U's property.

For example, in the example shown in FIG. 11, the user U identified by the user ID "U1" moves along the "location history #1", searches along the "search history #1", It indicates that the content was browsed according to the "Purchase history #1", the predetermined product etc. was purchased at the predetermined store etc. according to the "Purchase history #1", and the content was posted according to the "Posting history".

Here, in the example shown in FIG. 11, abstract history such as "U1", "location history #1", "search history #1", "browsing history #1", "purchase history #1", and "posting history #1" "U1", "location history #1", "search history #1", "browsing history #1", "purchase history #1" and "posting history #1" , information such as specific character strings and numerical values are stored.

Note that the history information database 122 may store various types of information, not limited to the above, depending on the purpose. For example, the history information database 122 may store the user U's usage history of a predetermined service. In addition, the history information database 122 may store the user U's store visit history, facility visit history, and the like. Also, the history information database 122 may store a payment history of payment (electronic payment) using the client terminal 10 of the user U, and the like.

(Anonymous information database 123)
The anonymous information database 123 stores various types of information regarding user U's anonymous information (vector data). FIG. 12 is a diagram showing an example of the anonymous information database 123. As shown in FIG. In the example shown in FIG. 12, the anonymous information database 123 has items such as "common ID", "company ID", "company", "vector", and "inference".

"Common ID" indicates a common customer ID in the data area. "Company ID" indicates a unique customer ID for each company. "Company" indicates the company that provided the anonymous information (vector data).

"Vector" indicates a vector value obtained by anonymizing (vectorizing) customer data for each company. A "vector" may be multi-dimensional. That is, a "vector" may be a set of vector values.

Also, "inference" indicates the result of inference based on vectors. For example, the “inference” may be a product purchase probability list (recommendation list) of each customer of each company. Note that "inference" is not limited to the result of inference for each company (for each company ID), but also includes the result of inference common to all companies (common ID).

For example, in the example shown in FIG. 12, the user U identified by the common ID "U1" is managed by the customer data of "A company" by the company ID "U1A", and the customer data is anonymized (vector It shows that the "inference A1" can be obtained based on the "vector A1" converted into

Here, in the example shown in FIG. 12, abstract values such as "U1", "U1A", "A company", "vector A1" and "inference A1" are used. ”, “Company A”, “Vector A1”, and “Inference A1” store information such as specific character strings and numerical values. Hereinafter, abstract values may also be illustrated in diagrams relating to other information.

It should be noted that the anonymous information database 123 may store various types of information, not limited to the above, depending on the purpose. For example, the anonymous information database 123 may store information about AI on the client side and the mother side. The anonymous information database 123 may also store information about AI learning and inference techniques.

(control unit 130)
Returning to FIG. 9, the description is continued. The control unit 130 is a controller, and for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), etc., controls the mother server 100. Various programs (corresponding to an example of an information processing program) stored in an internal storage device are executed by using a storage area such as a RAM as a work area. In the example illustrated in FIG. 9 , the control unit 130 has an acquisition unit 131 , a learning unit 132 , an estimation unit 133 and a provision unit 134 .

(Acquisition unit 131)
The acquisition unit 131 acquires a search query input by the user U. For example, the acquisition unit 131 acquires the search query via the communication unit 110 when the user U inputs a search query into a search engine or the like and performs a keyword search.

The acquisition unit 131 also acquires user information about the user U via the communication unit 110 . For example, the acquisition unit 131 acquires, from the client terminal 10 of the user U, identification information (user ID, etc.) indicating the user U, location information of the user U, attribute information of the user U, and the like. Further, the acquisition unit 131 may acquire identification information indicating the user U, attribute information of the user U, and the like when the user U is registered as a user. Acquisition unit 131 then registers the user information in user information database 121 of storage unit 120 .

In addition, the acquisition unit 131 acquires various types of history information (log data) indicating actions of the user U via the communication unit 110 . For example, the acquisition unit 131 acquires various types of history information indicating actions of the user U from the client terminal 10 of the user U or from various servers based on the user ID or the like. The acquisition unit 131 then registers various types of history information in the history information database 122 of the storage unit 120 .

Also, the acquisition unit 131 acquires anonymous information obtained by anonymizing (vectorizing) customer data from the client terminals 10 of each company participating in the data zone via the communication unit 110 . For example, the acquisition unit 131 acquires anonymous information obtained by anonymizing (vectorizing) customer data by the client-side AI from the client terminal 10 of each company via the communication unit 110 .

At this time, the acquisition unit 131 acquires anonymous information obtained by anonymizing (vectorizing) customer data from each of the client terminals 10 of a plurality of companies in different industries. Competitors (other companies in the same industry) may be included in the plurality of companies.

Anonymous information is multidimensional vector data obtained by vectorizing customer data, and is irreversible information for which there is no means for restoring customer data. In other words, anonymous information is not encrypted customer data and cannot be decrypted.

(learning unit 132)
The learning unit 132 performs AI machine learning using anonymous information from the client terminal 10 of each company. At this time, the learning unit 132 anonymizes (vectorizes) customer data (mother-side data) independently collected by the mother server 100 separately from each company to the anonymous information from the client terminal 10 of each company. AI may be machine-learned by adding information.

Also, the learning unit 132 sets the purpose of a specific company by using, as correct data, data corresponding to the purpose among the anonymous information from the client terminal 10 of each company during AI machine learning. At this time, the learning unit 132 selects the data corresponding to the purpose among the anonymous information from the client terminal 10 of each company and the customer data collected independently by the mother server 100 separately from each company as correct data. You may make it set the objective of a specific company. In addition, if the mother-side AI can also anonymize (anonymize (vectorize)) customer data and convert it into anonymous information in the same way as the client-side AI, the customer data collected independently by the mother server 100 can be directly used by the mother-side AI. can be entered in

(Estimation unit 133)
The estimating unit 133 estimates a customer who matches the purpose of a specific company among a plurality of companies by AI that has undergone machine learning by the learning unit 132 .

It should be noted that the estimation unit 133 more accurately estimates a customer who meets the purpose of a specific company among a plurality of companies as the number of anonymous information items from the specific company increases. Conversely, the smaller the number of anonymous information from a specific company, the lower the accuracy of estimating customers who meet the purpose of the specific company.

Also, the estimation unit 133 estimates a customer who has a high probability of purchasing a specific product from a specific company among a plurality of companies, and generates a product purchase probability list in which the customers are listed.

(Providing unit 134)
The providing unit 134 provides client-side AI for anonymizing (vectorizing) customer data to the client terminals 10 of each of the plurality of companies via the communication unit 110 .

Also, the providing unit 134 provides information based on the estimation result by the estimating unit 133 via the communication unit 110 . For example, the providing unit 134 provides the client terminal 10 of the specific company with a product purchase probability list in which customers with a high probability of purchasing the specific product of the specific company are listed via the communication unit 110 . As a result, a specific company (client side) distributes advertisements and coupons to customers listed on the list from the mother side, and as a result acquires new customers. By providing information to the mother side, the accuracy of AI's inference will be further improved.

Also, the providing unit 134 distributes advertisements and coupons relating to specific products of specific companies to customers listed on the product purchase probability list via the communication unit 110 . As a result, the specific company (client side) acquires new customers and provides anonymous information on the new customers to the mother side, further improving the accuracy of AI's inference.

[3. Processing procedure]
Next, a processing procedure by the mother server 100 according to the embodiment will be described with reference to FIG. 13 . FIG. 13 is a flowchart illustrating a processing procedure according to the embodiment; Note that the processing procedure described below is repeatedly executed by the control unit 130 of the mother server 100 .

As shown in FIG. 13, the providing unit 134 provides client-side AI for anonymizing (vectorizing) customer data to the client terminals 10 of each of a plurality of companies via the communication unit 110 (step S101). ).

Subsequently, the acquisition unit 131 of the mother server 100 acquires anonymous information obtained by anonymizing (vectorizing) customer data using the client-side AI from the client terminals 10 of each company participating in the data area via the communication unit 110. (step S102). Each company participating in the data sphere includes competitors and companies in other industries.

Subsequently, the learning unit 132 of the mother server 100 performs AI machine learning using anonymous information from the client terminals 10 of each company (step S103). At this time, the learning unit 132 anonymizes (vectorizes) anonymous information from the client terminal 10 of each company and customer data (mother-side data) independently collected by the mother server 100 separately from each company. AI machine learning using information doors.

Next, during AI machine learning, the learning unit 132 of the mother server 100 adds data corresponding to the purpose of the anonymous information from the client terminals 10 of each company to the mother server 100 independently of each company. The purpose of the specific company is set by using the collected customer data as correct data (step S104). It is optional whether or not to add customer data on the mother side. Of course, there is no problem even if only the data corresponding to the purpose among the anonymous information from the client terminal 10 of each company is used as the correct data.

Subsequently, the estimating unit 133 of the mother server 100 estimates a customer who matches the purpose of a specific company among a plurality of companies by AI that has undergone machine learning by the learning unit 132 (step S105). It should be noted that the estimation unit 133 more accurately estimates a customer who meets the purpose of a specific company among a plurality of companies as the number of anonymous information items from the specific company increases.

Subsequently, the estimation unit 133 of the mother server 100 generates a list of customers who meet the purpose of the specific company (step S106). For example, the estimation unit 133 generates a product purchase probability list in which customers with a high probability of purchasing a specific product of a specific company are listed.

Subsequently, the provision unit 134 of the mother server 100 provides the client terminal 10 of the specific company with a list of customers who meet the purpose of the specific company via the communication unit 110 (step S107). For example, the providing unit 134 provides the client terminal 10 of the specific company with a product purchase probability list in which customers with a high probability of purchasing the specific product of the specific company are listed via the communication unit 110 .

Subsequently, the providing unit 134 of the mother server 100 provides information in line with the purpose of the specific company to customers who meet the purpose of the specific company via the communication unit 110 (step S108). For example, the provision unit 134 distributes advertisements and coupons regarding specific products to customers listed on the product purchase probability list via the communication unit 110 .

[4. Modification]
The client terminal 10 and the mother server 100 described above may be implemented in various different forms other than the above embodiments. So, below, the modification of embodiment is demonstrated.

(standalone)
In the above embodiment, part or all of the processing executed by the mother server 100 may actually be executed by the client terminal 10 . For example, the processing may be completed stand-alone (by the client terminal 10 alone). In this case, it is assumed that the client terminal 10 has the functions of the mother server 100 in the above embodiment. In addition, in the above embodiment, the client terminal 10 cooperates with the mother server 100, so from the user U's point of view, it appears that the client terminal 10 is executing the processing of the mother server 100 as well. That is, from another point of view, it can be said that the client terminal 10 includes the mother server 100 . Also, the mother server 100 may be one of the plurality of client terminals 10 .

(basis vector)
In the above-described embodiment, in order to further improve the anonymity of the anonymous information, the mother server 100 predetermines the base vector and collects the difference data from the base vector as anonymous information from the client terminal 10. may In this case, only vector difference data is transmitted from the client terminal 10 to the mother server 100 .

(The format of anonymous information changes for each company)
Also, in the above embodiment, the format of the anonymous information may be changed for each company. For example, when the anonymous information is multi-dimensional vector data, the vector data of company A and the vector data of company B do not need to be arranged in the same dimension. Even if the information has the same content, the vector data of Company A and the vector data of Company B may be arranged in different dimensions. It is sufficient if only the AI on the mother side can recognize which data is arranged in which dimension of the multidimensional vector data of each company. In this case, the AI on the mother side may match (match) the vector data of each company prior to learning and inference. For example, vector data of each company is converted into vector data in a common format for learning and inference.

(AI learning on the client side)
In the above embodiment, the mother server 100 distributes AI to the client terminals 10 of each company. The mother server 100 may be provided with information on the specifications of the generated AI (vector conversion rules, etc.) or the AI itself (vector conversion model, etc.).

For example, the client terminal 10 uses customer data and vector values as a data set to construct a learning model using a machine learning technique using a neural network. Then, the client terminal 10 inputs customer data into the constructed learning model and converts the customer data into a multidimensional real-value vector. That is, the client terminal 10 is generated through machine learning using, for example, a plurality of fully connected layers, VAE (Variational Autoencoder), RNN (Recurrent Neural Network), LSTM (Long short-term memory), Transformer, BERT, and the like. Customer data is input to the vector conversion model that is used, and anonymous information obtained by anonymizing (vectorizing) the customer data is obtained as output.

Note that the RNN and LSTM may be neural networks based on an attention mechanism. Attention can handle data such as sentences in which the sequence is important. Also, the client terminal 10 may use a similar natural language processing model. By generating vector values from customer data using such a model, it is possible to generate vector values that place more importance on the order of information.

In response to a request from the mother server 100 or voluntarily, the client terminal 10 of each company transmits anonymous information obtained by anonymizing (vectorizing) customer data to the mother server 100, or at any time. Then, the client terminal 10 of each company transmits difference data (or the model itself) of the learned vector conversion model to the mother server 100 .

Also, the client terminal 10 of each company may generate a vector conversion model for converting customer data into a vector using a method such as federated learning. Federated learning is a machine learning method that jointly builds a learning model without exposing the data to outside the company. For example, in federated learning, the client terminal 10 of each company performs machine learning on the vector conversion model using customer data as learning data, and the difference data between the original model and the model after learning is transferred to the mother server 100. Send to The mother server 100 receives the difference data from each client terminal 10, integrates and learns them, and generates a common model. Then, the mother server 100 provides each client terminal 10 with the parameters of the generated common model. In this way, by using federated learning, it is possible to perform learning without providing customer data, which is learning data, from the client side to the server side, so it is possible to consider the customer's personal information and privacy. can be done.

[5. effect〕
As described above, the information processing apparatus (mother server 100) according to the present application includes an acquisition unit that acquires anonymous information obtained by vectorizing customer data from each of a plurality of companies, and an AI using anonymous information from each company. a learning unit that performs machine learning, an estimation unit that estimates customers who match the purpose of a specific company among multiple companies by AI that has undergone machine learning, and a provision unit that provides information based on the estimation results of the estimation unit. Prepare.

Anonymous information is multidimensional vector data obtained by vectorizing customer data, and is irreversible information for which there is no means for restoring customer data.

Further, the estimating unit more accurately estimates the customer who matches the purpose of the specific company as the number of anonymous information items from the specific company increases among the plurality of companies.

Also, the providing unit provides client-side AI for vectorizing customer data to each of the plurality of companies. Then, the acquisition unit acquires anonymous information obtained by vectorizing customer data from each of the plurality of companies using the client-side AI.

In addition, the learning unit performs AI machine learning by adding anonymous information obtained by vectorizing customer data independently collected by the information processing device independently of each company to the anonymous information from each company.

In addition, the learning unit sets the purpose of a specific company by adding customer data independently collected by the information processing device separately from each company to the data corresponding to the purpose among the anonymous information from each company, and sets the purpose of AI. Do machine learning.

In addition, the estimation unit estimates a customer who has a high probability of purchasing a specific product from a specific company among a plurality of companies, and generates a product purchase probability list in which the customers are listed. Then, the providing unit provides the product purchase probability list to the specific company.

In addition, the providing unit distributes advertisements and coupons relating to specific products to customers listed on the product purchase probability list.

Also, the acquisition unit acquires anonymous information obtained by vectorizing customer data from each of a plurality of companies in different industries.

The information processing apparatus according to the present application can provide a mechanism that makes data area member companies want to output their own data by any one or combination of the above-described processes. Of course, the security can be further improved by encrypting the vectorized anonymous information. As a result, each company can improve data area cooperation and targeting accuracy without disclosing raw personal information. In addition, since there is a mechanism that prevents free riding, the accuracy as a whole improves synergistically. In addition, when viewed from the perspective of each company, even if it is in the data zone, other companies cannot see their company's data and situation (for example, from the perspective of company A, how many customers have visited their own store, etc. unknown to the company). In addition, if the mother side is developing a large number of services (portal site, etc.), it can also be used for mass marketing such as advertisement distribution and SNS.

[6. Hardware configuration]
Also, the client terminal 10 and the mother server 100 according to the above-described embodiments are implemented by a computer 1000 configured as shown in FIG. 14, for example. The mother server 100 will be described below as an example. FIG. 14 is a diagram illustrating an example of a hardware configuration; The computer 1000 is connected to an output device 1010 and an input device 1020, and an arithmetic device 1030, a primary storage device 1040, a secondary storage device 1050, an output I/F (Interface) 1060, an input I/F 1070, and a network I/F 1080 are buses. It has a form connected by 1090.

The arithmetic device 1030 operates based on programs stored in the primary storage device 1040 and the secondary storage device 1050, programs read from the input device 1020, and the like, and executes various processes. The arithmetic unit 1030 is implemented by, for example, a CPU (Central Processing Unit), GPU, TPU, MPU (Micro Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), or the like.

The primary storage device 1040 is a memory device such as a RAM (Random Access Memory) that temporarily stores data used for various calculations by the arithmetic device 1030 . The secondary storage device 1050 is a storage device in which data used for various calculations by the arithmetic device 1030 and various databases are registered. State Drive), flash memory, or the like. The secondary storage device 1050 may be an internal storage or an external storage. Also, the secondary storage device 1050 may be a removable storage medium such as a USB (Universal Serial Bus) memory or an SD (Secure Digital) memory card. Also, the secondary storage device 1050 may be a cloud storage (online storage), a NAS (Network Attached Storage), a file server, or the like.

The output I/F 1060 is an interface for transmitting information to be output to the output device 1010 that outputs various information such as a display, a projector, and a printer. (Digital Visual Interface), HDMI (registered trademark) (High Definition Multimedia Interface), and other standardized connectors. Also, the input I/F 1070 is an interface for receiving information from various input devices 1020 such as a mouse, keyboard, keypad, buttons, scanner, etc., and is realized by, for example, USB.

Also, the output I/F 1060 and the input I/F 1070 may be wirelessly connected to the output device 1010 and the input device 1020, respectively. That is, the output device 1010 and the input device 1020 may be wireless devices.

Also, the output device 1010 and the input device 1020 may be integrated like a touch panel. In this case, the output I/F 1060 and the input I/F 1070 may also be integrated as an input/output I/F.

Note that the input device 1020 includes, for example, optical recording media such as CDs (Compact Discs), DVDs (Digital Versatile Discs), PDs (Phase change rewritable discs), magneto-optical recording media such as MOs (Magneto-Optical discs), and tapes. It may be a device that reads information from a medium, a magnetic recording medium, a semiconductor memory, or the like.

Network I/F 1080 receives data from other devices via network N and sends the data to arithmetic device 1030, and also transmits data generated by arithmetic device 1030 via network N to other devices.

Arithmetic device 1030 controls output device 1010 and input device 1020 via output I/F 1060 and input I/F 1070 . For example, arithmetic device 1030 loads a program from input device 1020 or secondary storage device 1050 onto primary storage device 1040 and executes the loaded program.

For example, when the computer 1000 functions as the mother server 100 , the computing device 1030 of the computer 1000 implements the functions of the control unit 130 by executing programs loaded on the primary storage device 1040 . Further, arithmetic device 1030 of computer 1000 may load a program acquired from another device via network I/F 1080 onto primary storage device 1040 and execute the loaded program. Further, the arithmetic unit 1030 of the computer 1000 may cooperate with another device via the network I/F 1080, and call functions, data, etc. of the program from another program of the other device for use.

[7. others〕
Although the embodiments of the present application have been described above, the present invention is not limited by the contents of these embodiments. In addition, the components described above include those that can be easily assumed by those skilled in the art, those that are substantially the same, and those within the so-called equivalent range. Furthermore, the components described above can be combined as appropriate. Furthermore, various omissions, replacements, or modifications of components can be made without departing from the gist of the above-described embodiments.

Further, among the processes described in the above embodiments, all or part of the processes described as being automatically performed can be manually performed, or the processes described as being performed manually can be performed manually. All or part of this can also be done automatically by known methods. In addition, information including processing procedures, specific names, various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each drawing is not limited to the illustrated information.

Also, each component of each device illustrated is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution and integration of each device is not limited to the illustrated one, and all or part of them can be functionally or physically distributed and integrated in arbitrary units according to various loads and usage conditions. Can be integrated and configured.

For example, the above-described mother server 100 may be realized by a plurality of server computers, and some functions may be realized by calling an external platform or the like using an API (Application Programming Interface), network computing, or the like. Flexible to change.

Also, the above-described embodiments and modifications can be appropriately combined within a range that does not contradict the processing content.

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

1 information processing system 10 client terminal 100 mother server 110 communication unit 120 storage unit 121 user information database 122 history information database 123 anonymous information database 130 control unit 131 acquisition unit 132 learning unit 133 estimation unit 134 provision unit

Claims (11)

an acquisition unit that acquires anonymous information obtained by vectorizing customer data from each of a plurality of companies;
A learning unit that performs AI machine learning using anonymous information from each company;
an estimating unit that estimates a customer who meets the purpose of a specific company among the plurality of companies by the AI that has undergone machine learning;
and a providing unit that provides information based on the result of estimation by the estimating unit. The information processing apparatus according to claim 1, wherein the anonymous information is multidimensional vector data obtained by vectorizing the customer data, and is irreversible information for which there is no means for restoring the customer data. 2. The estimating unit, among the plurality of companies, more accurately estimates the customer who meets the purpose of the specific company as the number of anonymous information from the specific company increases. 3. The information processing device according to 2. The providing unit provides client-side AI for vectorizing the customer data to each of the plurality of companies,
The information processing apparatus according to any one of claims 1 to 3, wherein the acquisition unit acquires anonymous information obtained by vectorizing customer data from each of the plurality of companies by the client-side AI. . The learning unit performs machine learning of the AI by adding anonymous information obtained by vectorizing customer data independently collected by the information processing device separately from each company to anonymous information from each company. The information processing apparatus according to any one of claims 1 to 4. During the machine learning of the AI, the learning unit adds customer data independently collected by the information processing device separately from the companies to the data corresponding to the purpose among the anonymous information from the companies. 6. The information processing apparatus according to claim 5, wherein the data is used as correct data to set the purpose of the specific company. The estimating unit estimates a customer with a high probability of purchasing a specific product from the specific company among the plurality of companies, and generates a product purchase probability list listing the customers,
The information processing apparatus according to any one of claims 1 to 6, wherein the providing unit provides the product purchase probability list to the specific company. 8. The information processing apparatus according to claim 7, wherein said providing unit distributes advertisements and coupons relating to said specific product to customers listed on said product purchase probability list. The information processing apparatus according to any one of claims 1 to 8, wherein the acquisition unit acquires anonymous information obtained by vectorizing customer data from each of a plurality of companies in different industries. An information processing method executed by an information processing device,
an acquisition step of acquiring anonymous information obtained by vectorizing customer data from each of a plurality of companies;
A learning step of AI machine learning using anonymous information from each company;
an estimation step of estimating a customer who meets the purpose of a specific company among the plurality of companies by the AI that has undergone machine learning;
and a provision step of providing information based on the estimation result of the estimation step. an acquisition procedure for acquiring anonymous information obtained by vectorizing customer data from each of a plurality of companies;
A learning procedure for AI machine learning using anonymous information from each company;
an estimation procedure for estimating a customer who meets the purpose of a specific company among the plurality of companies by the AI that has undergone machine learning;
An information processing program for causing a computer to execute a provision procedure for providing information based on the estimation results of the estimation procedure.

Images