Recommendation Systems

Abstract

To provide a recommendation system that can generate recommendations that satisfy users' needs with high accuracy.SOLUTION: A recommendation system 10 of the present invention includes: a first database D1 that stores personal data about a user's various attributes, and metadata; a second database D2 that stores data provided by an information provider; a server 1 that performs calculation processing based on the data stored in the first and second databases D1 and D2; a user terminal 2 of the user and an information provider terminal 3 of the information provider who communicate with the server 1 through a network NT; and a behavior prediction device 4.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a recommendation system that recommends and presents information that a user truly or potentially desires, based on various pieces of information about the user. [Background technology]

[0002] Traditionally, in order to understand the various needs of users, personal data such as website browsing history, location information, and purchase history has been collected. This type of personal data is used by various businesses for recommendation services and targeted advertising.

[0003] For example, Patent Document 1 discloses a system that notifies users of facility information recommendations. In the system described in Patent Document 1, facilities for which the user should be encouraged to make a reservation are determined based on the user's calendar information and attribute information, and a predetermined recommendation regarding the facility is notified to the user's terminal. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-174166 A Summary of the Invention [Problem to be solved by the invention]

[0005] As mentioned above, traditionally, each business independently collected personal data and used that data for recommendations. However, in recent years, from the perspective of personal information protection, as represented by the Personal Information Protection Act, restrictions have been imposed on the use of personal data independently collected by each business, and it is not expected that personal data collected by one business will be used by other businesses. For this reason, when the same individual uses services provided by multiple businesses, there has been a problem in that personal data is dispersed and cannot be linked for use.

[0006] In addition, because no business can collect personal data comprehensively, the personal data tends to be partial and related to the individual, and recommendations generated based on partial personal data are often uninteresting or unnecessary for the user. Furthermore, there is also the problem that distributing a lot of information that does not meet the user's needs increases the costs of distributing information for each business. [Means for solving the problem]

[0007] The technical objective of the present invention is to provide a recommendation system that can consider and improve the current situation as described above, and generate recommendations that meet user needs with a high degree of accuracy.

[0008] The recommendation system of the present invention includes a first database that stores personal data related to various attributes of a user and metadata associated therewith, a second database that stores data provided by an information provider, a server that performs calculations based on the data stored in the first and second databases, a user terminal of the user and an information provider terminal of the information provider that communicate with the server via a network, and a behavior prediction device. When the user terminal operated by the user communicates with the server for the first time, the user terminal is operated according to a personal data acquisition program driven by the server, and the personal data and metadata input by the user are transmitted to the server together with a specific code of the user. temporarily stored in the first database; The initial registration process for the user is executed by the server, and the user terminal operated by the user after the initial registration process communicates with the server. When the user terminal uses a program driven by the server, the server: Read from the first databaseThe behavior prediction device performs a calculation based on the personal data and the metadata to predict the next behavior of the user, performs a calculation to match the predicted data with the provided data registered in the second database, extracts the provided data optimal for the user, and notifies the user terminal to execute a recommendation service. When an operation corresponding to the recommendation of the provided data displayed is performed through the user terminal, the server attaches the behavioral data of the user with respect to the displayed provided data and the prediction data of the behavior prediction device to the personal data as the metadata, stores and accumulates them in the user terminal, and the personal data and the metadata temporarily stored in the first database are later erased from the first database. That is it.

[0010] In the recommendation system of the present invention, after an action corresponding to the displayed recommendation of the provided data is completed, emotion data including at least one of an evaluation of the information provider and an evaluation of a service provided by the information provider is input to the user terminal, and the emotion data is attached to the personal data as the metadata, and the user terminal and accumulates the personal data and the metadata temporarily stored in the first database, and the personal data and the metadata are later deleted from the first database. This may be done. Effect of the Invention

[0012] According to the recommendation system of the present invention, the personal data and metadata of the user who uses the recommendation system are accumulated and repeated, including emotions and satisfaction levels, as the user accumulates experiences and actions, so that the accuracy of recommending optimal data to be provided to the user can be improved, and services that provide high satisfaction to the user can be provided. Even when the same individual uses services provided by multiple businesses and advertising providers, if the personal data and metadata are stored in the user's terminal, they will not be dispersed, and it will be possible to link them to and use services provided by multiple businesses and information providers.

[0013] In addition, by collecting the characteristics and habits of users' behavior and emotions, it is possible to use past behavior and emotions as a basis for judgment, and the results of decision-making based on this can be accumulated as personal data and metadata, which creates great value. For example, by being able to check records of past behavior and emotions, it can help users with their judgment criteria in cases where a similar decision is required.

[0014] In addition to displaying the right amount of timely and appropriate information by predicting user behavior, unnecessary information is almost never displayed, allowing users to selectively choose what information to use and shortening search time. Furthermore, from the information provider's perspective, it can solve the problem of high information and communication costs caused by the transmission of large amounts of unnecessary information, and it can also be effective in suppressing negative factors such as a decline in brand image. [Brief description of the drawings]

[0015] [Figure 1] FIG. 2 is a functional block diagram showing the configuration of a recommendation system according to a basic embodiment. [Diagram 2] FIG. 2 is a chart showing the operation of the recommendation system in the first embodiment. [Diagram 3] FIG. 11 is a chart showing the operation of the recommendation system in the second embodiment. [Figure 4] FIG. 1 is a conceptual diagram showing data relationships in a recommendation system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Next, a specific embodiment of the present invention will be described with reference to the drawings (FIGS. 1 to 4).

[0017] The recommendation system 10 of the basic embodiment includes a first database D1 that stores personal data related to various attributes of users and associated metadata, and a second database D2 that stores advertisement data, which is an example of provided data. In addition, a secretariat server 1 that performs calculations based on the various data stored in the first and second databases D1 and D2 is configured to be able to communicate with a user terminal 2 and an advertisement provider terminal 3 via a network NT (Internet line).

[0018] The secretariat server 1 is a server device operated by a company that provides a social network service (hereinafter referred to as SNS as appropriate). As an example, the secretariat server 1 provides various services to a user terminal 2 on which a client-side communication application that receives the SNS is installed. For example, the secretariat server 1 can provide services such as a chat-style message exchange service, a My Page function that makes public one's profile and avatar, a visit history reference function that allows one to refer to the history of users who have visited My Page, a diary function that allows one to easily update entries like a blog, and a recommendation service, which will be described later.

[0019] When the user terminal 2 operated by a user communicates with the secretariat server 1 for the first time, the user terminal 2 is first operated according to a personal data acquisition program driven by the secretariat server 1. Then, the personal data and metadata entered by the user are transmitted to the secretariat server 1 together with the user's identification code (user ID) and stored (temporarily stored) in the first database D1. By carrying out the above communication between the secretariat server 1 and the user terminal 2, the secretariat server 1 executes the initial registration process for the user.

[0020] The personal data in the basic embodiment includes, for example, demographic data on demographic attributes such as the user's gender, age, area of ​​residence, educational background, occupation, income, and family structure, geographic data on geographic attributes such as the user's range of movement and current location, and psychological data on the user's lifestyle, preferences, values, and purchasing motivations. The types of each attribute data can be set or changed as appropriate.

[0021] In particular, in the basic embodiment, attention is paid to data related to lifestyle, etc., among psychological attribute data, and by having the user input the user's daily habits, to-do list, regular plans such as regular meetings, and daily routines, this personal data is utilized to deliver advertisement data useful to the user in a timely manner. For example, advertisement data for a pillow that helps a good night's sleep will not be displayed during a meeting to the user terminal 2 of a user who suffers from insomnia, and messages from a woman met on a dating app will not be displayed to the user terminal 2 of a user who is spending time with his / her partner.

[0022] The personal data and metadata sent from the user terminal 2 to the secretariat server 1 may be all or part of the data stored in the terminal storage unit 5 of the user terminal 2. It is preferable that the personal data and metadata are temporarily stored in the first database D1 when recommending specific advertising data, and are deleted from the first database D1 after use, and are stored and accumulated in the terminal storage unit 5 of the user terminal 2.

[0023] With the above configuration, even if the same individual uses services provided by multiple businesses or advertising providers, as long as the personal data and metadata are stored in the terminal memory unit 5 of the user terminal 2, they will not be dispersed, and it will be possible to link the personal data and metadata to and use the services provided by multiple businesses or advertising providers (see Figure 4).

[0024] The terminal storage unit 5 may store personal data and metadata so that only the user or a person authenticated by the user can access them. For example, the terminal storage unit 5 may conceal the personal data and metadata (make it difficult to access the personal data and metadata). Examples of concealment include encryption and secret sharing.

[0025] When the advertisement provider terminal 3 operated by an advertisement provider communicates with the secretariat server 1, the advertisement provider terminal 3 is operated according to an advertisement registration program driven by the secretariat server 1. As a result, advertisement data from the advertisement provider is transmitted to the secretariat server 1 and stored (temporarily stored) in the second database D2. By carrying out such communication between the server 1 and the advertisement provider terminal 3, the advertisement data of each advertisement provider is registered in the secretariat server 1, and the secretariat server 1 can select and understand what advertisement data provided by each advertisement provider is useful to users by checking the second database D2.

[0026] Next, with reference to Figures 1 and 2, we will explain a first embodiment in which specific advertising data is recommended to a pair of matched users (hereinafter referred to as a pair) by collecting personal data and metadata of the pair and learning their characteristics.

[0027] When the user terminal 2 operated by the user after the initial registration process communicates with the secretariat server 1 and the users are matched using an SNS program driven by the secretariat server 1, the secretariat server 1 identifies the matched pair from the matching results. Here, the matching results are the results of users being matched because they have mutual liking for each other. The secretariat server 1 displays photos of the matched partner on each user terminal 2, and when it receives an operation indicating mutual liking (e.g., clicking, tapping, etc.), it presumes that there is mutual liking and identifies the users as a pair.

[0028] The secretariat server 1 is equipped with a behavior prediction device 4. The behavior prediction device 4 generates a teacher model and a customer journey (a process of a user's behavior, thoughts, emotions, etc.) from the personal data and metadata of a large number of users, and predicts the user's next behavior. It then accumulates data on cases where the user's behavior was as predicted or different, and corrects the behavior prediction.

[0029] Here, the teacher model is a model for predicting user behavior with as high accuracy as possible, and is not intended to improve the accuracy of user behavior. It is natural that individual users will behave differently from the teacher model, and the purpose of the teacher model is rather to better match each individual data and metadata to the user by accumulating different behavioral data and emotional data within the SNS.

[0030] Therefore, by collecting the characteristics and habits of the user's behavior and emotions, the past behavior and emotions can be used as judgment criteria, and the results of decision-making based on this can be stored as personal data and metadata in the terminal storage unit 5 of the user terminal 2, which creates great value. For example, the ability to check records of past behavior and emotions can help the user's judgment criteria in cases where a similar decision is required.

[0031] When the secretariat server 1 identifies a pair, it reads the personal data and metadata of each user constituting the pair from the terminal storage unit 5 of each user terminal 2 or from the first database D1, and performs calculations using the behavior prediction device 4 based on the personal data and metadata of each member of the pair, the teacher model, and the customer journey to predict the next behavior of the pair (or each user). Then, the secretariat server 1 performs calculations to match the predicted data with each piece of advertising data registered in the second database D2, extracts optimal advertising data for the pair, and notifies each user terminal 2 of the pair. In other words, a recommendation service is executed for each user terminal 2 of the pair.

[0032] As a result, each user terminal 2 of the pair displays the most suitable advertising data for the pair on the display screen. Therefore, each member of the pair can check the advertising data corresponding to the attribute data (particularly psychological attribute data) of the pair through the display of the user terminal 2 without searching, and if they like the content of the advertising data, they can purchase or reserve it, making it possible to provide a service that provides high satisfaction to users. Uninteresting and unnecessary advertising data is not shown to the user, improving convenience for the user.

[0033] When a purchase operation or reservation operation (e.g., a click or a tap) corresponding to the recommendation of the displayed advertising data is performed through one of the user terminals 2 of the pair, the secretariat server 1 stores and accumulates (also called updates or overwrites) the behavioral data of the user (pair) such as the purchase or reservation for the displayed advertising data and the prediction data of the behavior prediction device 4 in the terminal memory unit 5 of each user terminal 2 of the pair as metadata (ancillary data) for personal data.

[0034] The user's behavioral data and the prediction data of the behavior prediction device 4 may be stored (temporarily stored) in the first database D1 as metadata for personal data. The user's behavioral data and the prediction data of the behavior prediction device 4 are also used to update the teacher model and customer journey in the behavior prediction device 4. The secretariat server 1 and the advertisement provider terminal 3 may obtain the user's behavioral data and the prediction data of the behavior prediction device 4 to calculate the conversion rate of the displayed advertisement data.

[0035] After a purchase, reservation, or visit corresponding to a recommendation in the displayed advertising data is completed, evaluation data (emotion data) including at least one of an evaluation of the advertising provider and an evaluation of the service provided by the advertising provider is input to each user terminal 2 of the pair. The evaluation data is also stored and accumulated as metadata for personal data in the terminal storage unit 5 of each user terminal 2. After a purchase, reservation, or visit corresponding to a recommendation in the displayed advertising data is completed, the secretariat server 1 may receive the evaluation data through the user terminal 2 and store it (temporarily store it) in the first database D1 as metadata for personal data.

[0036] The collection of the personal data and associated metadata as described above and the learning of the characteristics are repeatedly performed and accumulated (updated and overwritten). This improves the accuracy of recommending optimal advertising data to the user, and provides a service that satisfies the user. The personal data and metadata can be temporarily stored in the first database D1, but it is preferable to store and accumulate them in the terminal storage unit 5 of the user terminal 2.

[0037] Next, with reference to FIG. 1 and FIG. 3, a second embodiment will be described in which specific advertising data in accordance with a life cycle of a user is recommended to the user by collecting personal data and metadata of the user and learning the characteristics of the user.

[0038] When the user terminal 2 operated by the user after the initial registration process communicates with the secretariat server 1 and the user terminal 2 uses an SNS program driven by the secretariat server 1, the secretariat server 1 reads the user's personal data and metadata from the terminal storage unit 5 of the user terminal 2 or the first database D1, and performs calculations based on the personal data, metadata, teacher model, and customer journey using the behavior prediction device 4 to predict the user's next behavior. Then, the secretariat server 1 performs calculations to match the predicted data with each advertisement data registered in the second database D2, extracts advertisement data optimal for the user, and notifies the corresponding user terminal 2. In other words, a recommendation service is executed for the corresponding user terminal 2.

[0039] For example, while the user is on the way home or on the move, the system predicts the user's next action (such as preparing dinner after returning home) and displays advertising data on the user terminal 2 about the nearest stores (sale data at supermarkets, etc.) that the user will be passing by.

[0040] This allows the user terminal 2 to display the most suitable advertising data for the user on the display screen. Therefore, the user can check the advertising data corresponding to the user's attribute data (particularly psychological attribute data) through the display on the user terminal 2 without searching, and if the content of the advertising data is to his / her liking, he / she can purchase or reserve it. Therefore, as in the case of the first embodiment, it is possible to provide a service that satisfies the user. Uninteresting and unnecessary advertising data is not shown to the user, and convenience for the user can be improved.

[0041] When a purchase operation or reservation operation (e.g., clicking or tapping) corresponding to the recommendation of the displayed advertising data is performed through the user terminal 2, the secretariat server 1 stores and accumulates (also called updates or overwrites) the user's behavioral data such as the purchase or reservation for the displayed advertising data and the prediction data of the behavior prediction device 4 in the terminal memory unit 5 of the user terminal 2 as metadata (ancillary data) for the personal data.

[0042] In this case as well, the user's behavioral data and the prediction data of the behavior prediction device 4 can be stored (temporarily stored) in the first database D1 as metadata for personal data. As in the first embodiment, the user's behavioral data and the prediction data of the behavior prediction device 4 are also used to update the teacher model and customer journey in the behavior prediction device 4. The secretariat server 1 and the advertisement provider terminal 3 may obtain the user's behavioral data and the prediction data of the behavior prediction device 4 to find the conversion rate of the displayed advertisement data.

[0043] After a purchase, reservation, or visit corresponding to a recommendation in the displayed advertising data is completed, evaluation data (emotion data) including at least one of an evaluation of the advertising provider and an evaluation of the service provided by the advertising provider is input to each user terminal 2 of the pair. The evaluation data is also stored and accumulated as metadata for personal data in the terminal storage unit 5 of each user terminal 2. After a purchase, reservation, or visit corresponding to a recommendation in the displayed advertising data is completed, the secretariat server 1 may receive the evaluation data through the user terminal 2 and store (temporarily store) it in the first database D1 as metadata for personal data.

[0044] The collection of the personal data and associated metadata as described above and the learning of the characteristics are repeatedly performed and accumulated (updated and overwritten). This improves the accuracy of recommending optimal advertising data to users, and provides services that are highly satisfying to users. In this case, the personal data and metadata can be temporarily stored in the first database D1, but it is preferable to store and accumulate them in the terminal storage unit 5 of the user terminal 2.

[0045] The configuration of each part in the present invention is not limited to the embodiment shown in the figure, and various modifications are possible within the scope of the present invention. For example, the provided data is not limited to advertisement data, and when the recommendation system of the present invention is used in an education business, problem data of various levels of difficulty depending on the learning progress, etc. can be used, and when the system is used in a job-type task management business, task data such as instructions from superiors, assignments, help from colleagues, meeting minutes, and orders from business partners can be used. [Explanation of symbols]

[0046] D1 First Database D2 Second Database NT Network 1. Secretariat Server 2 User terminal 3. Advertiser terminal 4. Behavioral Prediction Device 5. Terminal memory section 10 Recommendation Systems

Claims

1. a first database for storing personal data relating to various attributes of a user and metadata associated therewith; a second database for storing data provided by an information provider; a server for performing calculations based on data stored in the first and second databases; a user terminal of the user and an information provider terminal of the information provider, which communicate with the server via a network; and a behavior prediction device; When the user terminal operated by the user communicates with the server for the first time, the user terminal is operated according to a personal data acquisition program driven by the server, the personal data and the metadata input by the user are transmitted to the server together with the user's identification code and temporarily stored in the first database, and an initial registration process for the user is executed by the server; When the user terminal operated by the user after the initial registration process communicates with the server and the user terminal uses a program driven by the server, the server performs a calculation using the behavior prediction device based on the personal data and the metadata read from the first database to predict the next behavior of the user, performing a calculation for matching the predicted data with the provided data registered in the second database, extracting the provided data optimal for the user, and notifying the user terminal of the extracted data to execute a recommendation service; When an operation corresponding to the displayed recommendation of the provided data is performed through the user terminal, the server attaches the behavioral data of the user with respect to the displayed provided data and the prediction data of the behavior prediction device to the personal data as the metadata, and stores and accumulates them in the user terminal, and the personal data and the metadata temporarily stored in the first database are later erased from the first database. Recommendation system.

2. After the action corresponding to the displayed recommendation of the provided data is completed, emotional data including at least one of an evaluation of the information provider and an evaluation of the service provided by the information provider is input to the user terminal, the emotional data is attached to the personal data as the metadata, and is stored and accumulated in the user terminal, and the personal data and the metadata temporarily stored in the first database are later erased from the first database. A recommendation system according to claim 1.

Images