Matching system, learning device, matching method, and learning method

Abstract

To provide a matching system capable of performing matching between persons while improving accuracy.SOLUTION: An information processing device includes an acquisition unit that acquires vital data which is a measurement result of a vital sensor for each of a plurality of target users, and a matching processing unit that obtains a degree of matching between users on the basis of the acquired vital data of the plurality of users.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a matching system, a learning device, a matching method, and a learning method. [Background technology]

[0002] There is a system that measures the brain activity of two subjects and predicts the compatibility between the two subjects using the measurement results (for example, Patent Document 1). In such a system, to understand the brain activity of the subjects, images are taken using functional magnetic resonance imaging (fMRI), and the compatibility is predicted using the imaging results. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-77800 Summary of the Invention [Problem to be solved by the invention]

[0004] However, taking fMRI images to predict compatibility requires the preparation of specialized equipment and the visit to a location where such specialized equipment is installed (e.g., a medical institution, etc.). While electroencephalography (EEG) and near-infrared spectroscopy (NIRS) measurement devices can also be used to understand brain activity, these devices are often installed in limited locations, such as medical institutions, and are therefore difficult to access. Furthermore, when estimating compatibility without using such dedicated equipment, one possible approach is to have subjects answer various questions in a questionnaire, and then use the answers to conduct a personality diagnosis and estimate compatibility. However, questionnaire-based methods are subject to personal subjectivity, making it difficult to accurately capture the personality of an actual user. Therefore, even if matching is performed using the results of a questionnaire, the accuracy is not necessarily sufficient.

[0005] The present invention has been made in consideration of the above circumstances, and its purpose is to provide a matching system, learning device, matching method, and learning method that can match people with improved accuracy even when using simple equipment. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems, one aspect of the present invention is a matching system having an acquisition unit that acquires vital data, which is the measurement result of a vital sensor, for each of multiple target users, and a matching processing unit that determines the degree of matching between users based on the acquired vital data of the multiple users.

[0007] Another aspect of the present invention is a learning device that learns the relationship between emotion data estimated based on vital data acquired from a first user, emotion data estimated based on vital data acquired from a second user, and the degree of matching between the first user and the second user.

[0008] Another aspect of the present invention is a matching method executed by a computer, comprising: This is a matching method in which vital data, which is the measurement result of a vital sensor, is acquired for each of a plurality of target users, and the degree of matching between users is determined based on the acquired vital data of the plurality of users.

[0009] Another aspect of the present invention is a learning method for learning the relationship between emotion data estimated based on vital data acquired from a first user, emotion data estimated based on vital data acquired from a second user, and the degree of matching between the first user and the second user. [Effects of the Invention]

[0010] As described above, according to the present invention, matching is performed using vital data, so that person-to-person matching can be performed using simple equipment, the influence of personal subjectivity is reduced, and accuracy is improved. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a schematic block diagram showing the configuration of a matching system S according to an embodiment of the present invention. [Figure 2] 2 is a functional block diagram showing the general functions of a matching server 40. FIG. [Figure 3] 10 is a flowchart illustrating a flow of using emotion estimation data. [Figure 4] 1 is a conceptual diagram illustrating the flow of operations of the matching system S. FIG. [Figure 5] 10 is a flowchart illustrating the flow of processing in the execution phase. [Figure 6] FIG. 10 is a diagram showing the relationship between emotion data of multiple users who are the target of matching. [Figure 7] 10 is a flowchart illustrating the operation of matching server 40 in the execution phase. [Figure 8] FIG. 10 is a diagram showing emotion data obtained based on a plurality of indices from vital data. DETAILED DESCRIPTION OF THE INVENTION

[0012] A matching system according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of a matching system S according to an embodiment of the present invention. The matching system S includes a plurality of terminal devices 10 (terminal device 10a, terminal device 10b, ... terminal device 10n), a plurality of vital sensors 15 (vital sensor 15a, vital sensor 15b, ... vital sensor 15n), a service server 20, a learning device 30, a matching server 40, and a network NW.

[0013] The multiple terminal devices 10 (terminal device 10a, terminal device 10b, ... terminal device 10n) are used by different users, and may be, for example, any of smartphones, tablets, personal computers, etc. Each of the multiple terminal devices 10 is communicably connected to a network NW, and communicates with other devices connected to the network NW. The multiple vital sensors 15 (vital sensor 15a, vital sensor 15b, ... vital sensor 15n) are each used by a different user and perform biometric sensing of the user's pulse, breathing, etc. from the movements of the user's body surface, generating vital data representing the sensing results. The vital data is data in which values ​​representing measurement results are arranged in chronological order. The vital sensors 15 may be contact sensors that sense biometrics by contacting a part of the user's body, or non-contact sensors that sense biometrics without contacting the user. If the vital sensor 15 is non-contact, it senses biometrics by, for example, irradiating the user to be measured with microwaves using a Doppler sensor and detecting the reflected waves. The vital sensor 15 can perform measurements from a position, for example, about 10 cm to several meters away from the user to be measured.

[0014] The contact-type vital sensor can be, for example, an existing electronic device that is built into a wristwatch and sold. Such wristwatch-type vital sensors are widely available, and the number of users who already own them is increasing. Non-contact vital signs sensors are increasingly being installed in homes and businesses, creating an environment where they can be easily used. In this way, vital sensors, whether contact or non-contact, are becoming easier to use than devices that measure brain activity.

[0015] Here, the terminal device 10a and the vital sensor 15a are used, for example, by user A, the terminal device 10b and the vital sensor 15b are used, for example, by user B, and the terminal device 10n and the vital sensor 15n are used, for example, by user N. The number of combinations of terminal devices 10 and vital sensors 15 may be set for each user, and may be, for example, two or more. For example, vital sensor 15a can transmit vital data, which is a measurement result, to terminal device 10a by wirelessly communicating with terminal device 10a through pairing processing. Similarly, vital sensor 15b and terminal device 10b, and vital sensor 15n and terminal device 10n are connected to each other so as to be able to communicate wirelessly.

[0016] Here, if the vital sensor 15 is communicably connectable to the network NW, it may communicate with other devices connected to the network NW without going through the terminal device 10.

[0017] The service server 20 provides a matching service. For example, based on a request from a user, the matching server 40 determines the degree of matching between a first user and a second user, and transmits data representing the degree of matching to the terminal device 10 of the requesting user. Such a service may be provided to the terminal device 10 via a website published by the service server 20. A lower degree of matching indicates a lower degree of compatibility between the users, and a higher degree indicates a higher degree of compatibility between the users.

[0018] The learning device 30 performs learning using training data and generates a trained model. The learning method performed by the learning device 30 may be machine learning or deep learning. The learning device 30 is communicably connected to a network NW and transmits the generated trained model to a matching server 40 via the network NW.

[0019] The matching server 40 determines the degree of matching between multiple users to be evaluated based on the vital data obtained from the users' terminal devices 10 or the vital sensors 15. The matching server 40 may also determine the degree of matching between multiple users to be evaluated based on the emotions of the users estimated based on the vital data. Matching server 40 is communicably connected to service server 20 and learning device 30 via network NW. Matching server 40 may also be communicably connected to terminal device 10 via network NW.

[0020] The network NW may be the Internet, a LAN (Local Area Network), or a combination thereof.

[0021] FIG. 2 is a functional block diagram showing an outline of the functions of matching server 40. As shown in FIG. The matching server 40 includes a communication unit 401 , a storage unit 402 , a content providing unit 403 , an acquisition unit 404 , a scene dividing unit 405 , an emotion estimating unit 407 , and a matching processing unit 408 . The communication unit 401 communicates with external devices via the network NW.

[0022] The storage unit 402 stores various types of data. The memory unit 402 is configured by a storage medium, such as a hard disk drive (HDD), flash memory, electrically erasable programmable read-only memory (EEPROM), random access read / write memory (RAM), read-only memory (ROM), or any combination of these storage media. The storage unit 402 may be, for example, a nonvolatile memory.

[0023] For example, the storage unit 402 stores an emotion estimation data storage unit 4021 and a measurement target behavior storage unit 4022 . The emotion estimation data storage unit 4021 stores emotion estimation data that indicates the correspondence between vital data and emotions. The emotion estimation data may be an emotion estimation model, which will be described later. The measurement target behavior storage unit 4022 stores measurement target behavior data, which is data related to behaviors to be measured. The measurement target behavior data may be a menu of behaviors to be measured, or may be content to be measured.

[0024] The acquisition unit 404 acquires vital data, which is a measurement result of a vital sensor, for each of a plurality of target users. The vital data can divide a period in which the behavior of the measurement target is being performed into a plurality of different scenes. The acquiring unit 404 may acquire time-series vital data measured during a period in which the behavior determined as the measurement target is being performed.

[0025] The scene division unit 405 divides the vital data by scene. The scene division unit may also divide, by scene, time-series emotion data obtained from vital data for sections not divided by scene. The scene division unit 405 may divide boundaries between different scenes in the time-series data based on attribute data that indicate set scene boundaries in content, etc. The scene division unit 405 may also divide the time-series data of vital data or emotion data based on an operation input from an operator, such as an operator, specifying a division position via an input device (keyboard, mouse, touch panel, etc.).

[0026] The emotion estimation unit 407 estimates emotion data representing an emotion corresponding to the vital data from the vital data.

[0027] The matching processing unit 408 determines the degree of matching between users based on the acquired vital data of a plurality of users. The matching processing unit 408 may use vital data corresponding to the target section for matching among the vital data to find the degree of matching between users. The matching processing unit 408 uses vital data corresponding to a scene to be matched among the scenes to determine the degree of matching between users. Furthermore, matching processing unit 408 may estimate emotion data that represents emotions according to vital data, and determine the degree of matching between users based on the estimated emotion data. In addition, the matching processing unit 408 may use a trained model trained by the learning device 30 and input vital data of the user to be evaluated into the trained model to obtain the degree of matching.

[0028] The communication unit 401, content providing unit 403, acquisition unit 404, scene division unit 405, emotion estimation unit 407, and matching processing unit 408 of the matching server 40 may be configured by a processing unit such as a CPU (Central Processing Unit) or a dedicated electronic circuit.

[0029] FIG. 3 is a flow diagram illustrating the flow of using emotion estimation data. The emotion estimation data is generated for user U1, who can provide vital data used in generating the emotion estimation data. The user U1 is asked to perform a behavior defined as a measurement target, and vital data is generated by measuring the user U1 during the period in which the behavior is performed using a vital sensor 15 that measures the user U1. Furthermore, a questionnaire is conducted regarding the emotions felt while the user U1 was performing the behavior, and the results of the questionnaire are input via the terminal device 10. Emotion estimation data that associates the vital data of user U1 while performing the behavior defined as a measurement target with the emotions resulting from the questionnaire is stored in the emotion estimation data storage unit 4021. Here, the user to be measured may be a single user U1, but it is preferable to measure a large number of different users. Here, the relationship between the vital data of user U1 while performing the behavior specified as the measurement target and the emotions of user U1 ascertained through a questionnaire may be used as training data to have the learning device 30 learn the relationship between the vital data and emotions, thereby generating a trained model for estimating emotions from the vital data, and storing this as an emotion estimation model (emotion estimation data) in the emotion estimation data storage unit 4021. In this way, by using the emotion estimation model, emotion data can be estimated from trends in changes in vital data such as heart rate and pulse rate.

[0030] Here, the behavior defined as the measurement target may be content viewing, eating, conversation, work at a company, organization, or the like, and the behavior may be divisible into multiple scenes. For example, in the case of content viewing, the content may include at least one scene. The scenes may be used to classify the behavior into time intervals depending on the content of the behavior, or may be used to classify the behavior into time intervals regardless of the content. Content that includes at least one scene may be moving images such as movies, dramas, television programs, and commercials, still images, and music such as songs and musical instrument performances. For example, a movie may have a scene in which the main character talks with another character, a scene in which the main character travels in a vehicle, a sunset scene, etc. A single still image may be used as one scene. A song may have an intro, an verse, a bridge, a chorus, etc., each considered a scene, or may be divided into sections based on the elapsed time from the start of playback of the song (the first scene is from the start of playback t0 until just before time t1, and the second scene is from time t1 until just before time t2).

[0031] Furthermore, if the action is work at work, each of the tasks of creating an email, making a phone call to a client, reporting to a superior, receiving in-house training, light work, etc. may be considered a scene, or each of these tasks may be further subdivided into scenes. Also, if the action is a conversation, the scenes may be divided according to the elapsed time from the start of the conversation, such that the first scene is from the start of the conversation t0 to just before time t1, and the second scene is from time t1 to just before time t2.

[0032] To obtain an emotion estimation model for each scene, the time-series vital data acquired from user U1 may be divided into time segments corresponding to the scenes, and the relationship between the vital data (divided vital data) belonging to the same scene and the emotions obtained by conducting a questionnaire about the emotions in that scene may be learned. In this way, an emotion estimation model for each scene can be obtained.

[0033] After the emotion estimation data is stored in the emotion estimation data storage unit 4021 in this manner, when vital data of a certain user (for example, user U1 or user U2 different from user U1) is obtained, the emotion estimation unit 407 can obtain emotion data corresponding to the vital data by referring to the emotion estimation data based on this vital data. Here, the emotion estimation unit 407 may obtain emotion data by inputting the vital data to an emotion estimation model. Based on the emotion data thus obtained, the degree of matching between a plurality of users can be determined.

[0034] Next, the operation of the matching system S will be described. FIG. 4 is a conceptual diagram illustrating the flow of operations of the matching system S. The operation of the matching system S can be broadly divided into a "preparation phase" and an "execution phase."

[0035] Preparation Phase The preparation phase is a phase in which emotion estimation data (or emotion estimation model) is generated and stored in emotion estimation data storage unit 4021. Here, a plurality of users are targeted, and while the users are asked to perform actions determined as measurement targets (step S1), vital data is measured for each user using the vital sensor 15. By performing measurements using the vital sensor 15, it is possible to obtain vital data, which is time-series data that changes according to emotions. Measurement by the vital sensor 15 may be any of the following.

[0036] <Measurement method (1)> A plurality of subjects are asked to perform the same action (for example, the same task), and changes in emotions during the task are measured by the vital sensor 15.

[0037] <Measurement method (2)> A user is asked to view certain content (a movie, a drama, a commercial, etc.), and changes in emotions during the viewing are measured by the vital sensor 15.

[0038] When such measurements are taken using vital sensor 15, terminal device 10 acquires vital data from vital sensor 15 and also asks the user to enter answers to a questionnaire about emotions during the measurement period using vital sensor 15, and transmits the vital data and the emotions resulting from the questionnaire to learning device 30. Learning device 30 acquires data in which such vital data and emotions are associated from each of the multiple users (step S2) and stores the data in a storage device within learning device 30. Learning device 30 generates an emotion estimation model by learning using data in which vital data and emotions are associated, stored in a storage device, and stores the generated emotion estimation model in the storage device (step S3). Here, learning device 30 transmits the generated emotion estimation model to matching server 40. Matching server 40 stores the emotion estimation model transmitted from learning device 30 in emotion estimation data storage unit 4021.

[0039] Execution Phase In the execution phase, the acquisition unit 404 of the matching server 40 acquires vital data of each of the multiple users to be measured (step S5). The emotion estimation unit 407 uses the acquired vital data and emotion estimation data to read emotion data corresponding to the vital data, thereby estimating the emotion of each user (step S6). Alternatively, the emotion estimation unit 407 may input the acquired vital data into an emotion estimation model and estimate the emotion of each user by obtaining emotion data derived by the emotion estimation model. The matching processing unit 408 calculates the degree of matching between the multiple users based on the estimated emotion data (step S7).

[0040] Next, the execution phase will be further described. FIG. 5 is a flowchart illustrating the process flow in the execution phase. Here, the service server 20 extracts users Ua and Ub for whom the matching degree is to be determined, and issues instructions on what actions to take on the terminal device 10a used by user Ua and the terminal device 10b used by user Ub, along with a notification to perform measurements using the vital sensor 15. The service server 20 may extract users for whom the matching degree is to be determined according to predetermined conditions (random extraction, extraction according to user attributes, etc.), or may extract users when requests to determine the matching degree from users Ua and Ub are received from the terminal devices 10, respectively.

[0041] Once the target users are extracted, at least two users for whom the matching degree is to be obtained are asked to perform the same behavior as the behavior to be measured (step S10). For example, if the target users for whom the matching degree is to be obtained are user Ua and user Ub, user Ua and user Ub are each asked to perform the same behavior. The behavior may be such that the two users each view the same content as when the emotion estimation model was generated, or may be such that the two users perform the same light work as when the emotion estimation model was generated.

[0042] The vital sensor 15 measuring user Ua transmits vital data, which is the measurement result obtained by measuring the pulse, breathing, etc. of user Ua at predetermined time intervals while the behavior to be measured is being performed, together with user identification information, to the matching server 40 via the terminal device 10. Similarly, the vital sensor 15 measuring user Ub transmits vital data, which is the measurement result obtained by measuring the pulse, breathing, etc. of user Ub at predetermined time intervals while the behavior to be measured is being performed, together with user identification information, to the matching server 40 via the terminal device 10.

[0043] Matching server 40 stores the vital data of user Ua and the vital data of user Ub in storage unit 402 (step S20). When the vital data of user Ua and user Ub are obtained, the emotion estimation unit 407 of the matching server 40 estimates an emotion according to the vital data of user Ua and an emotion according to the vital data of user Ub (step S21). The matching processing unit 408 calculates the degree of matching based on the emotion estimated from the vital data of user Ua and the emotion estimated from the vital data of user Ub (step S22).

[0044] After determining the degree of matching, matching server 40 transmits the determined degree of matching to service server 20 (step S22). When the service server 20 receives the matching degree from the matching server 40, it transmits the matching degree to each terminal device 10 that is the source of the vital data (step S11). As a result, user Ua and user Ub receive the matching degree on the terminal device 10 they use, and display the matching degree on the display screen of the terminal device 10. For example, the display screen of each terminal device 10 displays the user name of the matching target and the matching degree, such as "The matching degree between user Ua and user Ub is XX." This allows user Ua and user Ub to understand their matching degree.

[0045] Next, the matching process will be further described. FIG. 6 is a diagram showing the relationship between emotion data of multiple users who are the target of matching. This emotional data is time-series data of emotional data estimated based on vital data measured by vital sensor 15 while user Ua and user Ub were each viewing content C from playback time t1, which is the time when playback of content C started, to playback time t4. Here, user Ua and user Ub may be in the same room and view video displayed on the display and sound output from a speaker of one playback device, or may view content C using playback devices installed in different rooms. User Ua and user Ub may also view the same content C at the same time, or on different days.

[0046] In Fig. 6, the top diagram shows the relationship between the time elapsed since the start of playback of content C and the scene. The middle diagram shows the waveform of emotion data based on vital data obtained from user Ua. The bottom diagram shows the waveform of emotion data based on vital data obtained from user Ub. In emotion data, values ​​on the positive side (upper side) of a reference value (for example, 0) represent positive emotions, and values ​​on the negative side (lower side) represent negative emotions. Starting from the start of playback, content C progresses through scenes SC1, SC2, and SC3 in that order. In scene SC1, the changes in the emotion data of users Ua and Ub are generally the same. In scene SC2, the emotion data of user Ua has times when it is negative, but the positive values ​​increase as scene SC2 progresses. In contrast, the emotion data of user Ub has a long period of negative values ​​in scene SC2. In scene SC3, the emotion data of user Ua has a period of negative values ​​in the first half and a period of positive values ​​in the second half. In contrast, the emotion data of user Ub has a period of positive values ​​in the first half and a period of negative values ​​in the second half.

[0047] The times of these scenes may be determined in advance for each different scene based on the playback start time of the content, and may be included in the content C as attribute data for the content C.

[0048] Next, the operation of matching server 40 will be described. FIG. 7 is a flowchart illustrating the operation of matching server 40 in the execution phase. Here, emotion estimation data (emotion estimation model) is generated in advance and stored in emotion estimation data storage unit 4021. When the content providing unit 403 of the matching server 40 receives the designation of the users Ua and Ub to be measured and a request for matching processing from the service server 20 via the communication unit 401, the content providing unit 403 selects the behavior to be measured (step S101). Here, for example, the content providing unit 403 selects "viewing content" as the behavior to be measured and selects the content to be viewed. Then, the content providing unit 403 delivers the content to be viewed to the terminal device 10a of the user Ua and the terminal device 10b of the user Ub, which are designated as the measurement targets (step S102).

[0049] When content is distributed to the terminal device 10a of the user Ua and the terminal device 10b of the user Ub, the user Ua plays and watches the content on the terminal device 10a while measuring vital data during the content viewing period using the vital sensor 15a. When the content playback ends, the terminal device 10a receives the vital data measured during the content playback period from the vital sensor 15a and transmits the received vital data to the matching server 40. Similarly, user Ub plays and watches content on terminal device 10b, while vital sensor 15b measures vital data during the content viewing period. When the content playback ends, terminal device 10b receives the vital data measured during the content playback period from vital sensor 15b and transmits the received vital data to matching server 40.

[0050] Acquisition unit 404 of matching server 40 acquires vital data from terminal device 10a and terminal device 10b (step S103), and stores the acquired vital data in storage unit 402.

[0051] When the vital data is stored, the scene dividing unit 405 divides the stored vital data into scenes. Here, the scene dividing unit 405 divides the time-series data of the vital data into scenes based on time intervals representing the scenes by referring to attribute data set in the content transmitted by the content providing unit 403. Here, the scene dividing unit 405 divides the vital data of user Ua into scenes, and also divides the vital data of user Ub into scenes (step S104).

[0052] The emotion estimation unit 407 estimates emotion by inputting the vital data divided for each scene into an emotion estimation model to obtain emotion data for each scene (step S105). Here, the emotion estimation unit 407 estimates emotion for each scene for each user to be measured.

[0053] Once the emotions are estimated, the matching processing unit 408 determines the degree of matching between the user Ua and the user Ub based on the emotion data of the user Ua and the emotion data of the user Ub that correspond to the same scene (step S106). The matching processing unit 408 transmits the determined degree of matching to the terminal device 10a and the terminal device 10b (step S107). Here, the matching processing unit 408 may transmit the determined degree of matching to the service server 20, and the service server 20 may transmit the degree of matching to the terminal device 10a and the terminal device 10b. Here, the degree of matching may be expressed numerically, such as "compatibility XX%" or "compatibility level XX." Alternatively, the degree of matching may be expressed by a character string indicating the level of the degree, such as "compatibility: very good" or "compatibility: not so good."

[0054] Next, another example of emotion data will be described. 6 above, the emotional data is described as obtaining an evaluation result of "positive (good: feeling happy)" or "negative (bad: not feeling happy)" based on a single index of "whether or not one feels happy" from vital data, but the emotional estimation unit 407 may estimate emotional data by obtaining evaluation results for multiple indexes from vital data. Then, the matching processing unit 408 may determine the degree of matching between users using emotional data for each of the multiple indexes obtained from the vital data.

[0055] 8 is a diagram showing emotion data obtained from vital data based on multiple indices. As an example, the diagram shows a case where a waveform 801 indicating a first index "happiness," a waveform 802 indicating a second index "sadness," and a waveform 803 indicating a third index "stress" are obtained from the vital data of user Ua. In this way, emotion data for a plurality of indices may be obtained from vital data, and the emotion data based on these indices may be used to determine the degree of matching between users.

[0056] Matching processing unit 408 can perform matching processing using emotion data obtained from one vital data item based on one or more indices, and determine the degree of matching. Here, the evaluation method in the matching process executed by the matching processing unit 408 may be, for example, any of the following methods (1) to (4). (1) The degree of matching is obtained by scoring the entire time-series data of vital data or emotional data without classifying them by scene. (2) Time-series vital data or emotional data are classified by scene, and the degree of matching is obtained by scoring each scene. (3) When scoring in (1) or (2) above, score the emotion data of two users (or three or more users) to be evaluated by obtaining correlations between the users and obtain the degree of matching. (4) When scoring in (3) above, score by obtaining correlation between users using emotion data based on one index from vital data or multiple indexes, and obtain the degree of matching.

[0057] Furthermore, the matching processing unit 408 may use the following scoring methods (11) to (13), for example. (11) Evaluation based on the frequency of the same emotion being expressed in the evaluation section (whole or scene) among the users being evaluated. This evaluation may be performed using either of the following:

[0058] (11-1) Evaluation based on the cumulative time spent expressing positive or negative emotions For example, the degree of matching is evaluated by calculating the cumulative value of the time during which a positive emotion is expressed, and then calculating the correlation between the users for the cumulative value. Here, the cumulative value of the time during which a positive (or negative) emotion is expressed may be calculated based on emotion data obtained from the entire vital data, and the correlation may be calculated. Alternatively, the cumulative value of the time during which a positive (or negative) emotion is expressed may be calculated based on emotion data for each scene, and the correlation may be calculated for each scene.

[0059] (11-2) Evaluation based on the integral value (area) of the area where emotions, which indicate good or bad, are expressed For example, the degree of matching is evaluated by finding the area where the emotion indicating good is expressed and finding the correlation between users based on that area. Here, the area showing good (or bad) emotions may be calculated based on emotion data obtained from the entire vital data, and the correlation may be calculated. Alternatively, the area showing good (or bad) emotions may be calculated based on emotion data for each scene, and the correlation may be calculated for each scene.

[0060] (12) In addition to the process of (11) above, we use the association of multiple emotions. This (12) can be used when multiple indicators are obtained from vital data. For example, if emotional data based on three indicators, "happiness," "sadness," and "stress," is obtained from vital data, it is possible to define what can be said when "happiness" and "stress" are positive values ​​and "sadness" is a negative value, and the degree of matching can be determined from the evaluation results obtained based on that definition. Here, when multiple indicators can be obtained from vital data, one of the values ​​provided is the ability to output results that give meaning to the emotions in the range where "sadness" and "stress" are positive and have large numerical values, and "joy" is negative and has a small numerical value. Furthermore, for example, if the emotional data for the same time period (scene) is a combination in which user Ua has a positive value for "joy" and user Ub has a positive value for "sadness," the compatibility between user Ua and user Ub may be derived based on such a combination.

[0061] (13) In addition to the process of (12) above, derive a more compatible personality from the personality derived from emotional data. In this (13), the matching processing unit 408 estimates the personality based on the emotion data estimated from the vital data. Then, the matching processing unit 408 uses the estimated personality to determine the degree of matching between users. The personality here refers to the characteristics of an individual user, and may be, for example, a personality estimated based on time-series data of emotion data for one or more indicators. Personality may be, for example, "meticulous," "easily bored," "optimistic," "sincere," "extroverted," "introverted," etc., and at least one of these may be estimated. Furthermore, in the process of (13), the matching processing unit 408 may use the lingering emotion and the individual's physical condition (such as the relationship between weather and mood), thereby enabling a more accurate estimation of the matching degree. For example, when an emotion is stirred, the matching degree may be calculated taking into account that the effect lasts for a certain period of time when the lingering emotion is used. Furthermore, for the individual's physical condition, the matching degree may be calculated taking into account the effects of poor physical condition, changes before and after exercise, etc.

[0062] <Utilizing the degree of matching> Next, an example of utilizing the matching degree obtained as described above will be described. <Example of use: Matchmaking service> In recent years, the use of services that support marriage activities (hereinafter sometimes referred to as "konkatsu"), which are activities to find a marriage partner, has been increasing. Among marriage-hunting services, there are services that allow users to select and be introduced to potential marriage partners. These candidates are often selected based on annual income, occupation, age, etc. When users actually meet a candidate selected in this way, even if they meet certain criteria such as annual income, they may find it difficult to develop a deep relationship due to differences in personality and compatibility.

[0063] In addition to the method using conditions, there is also a method of conducting a personality diagnosis in the form of a questionnaire and extracting potential partners using the personality based on the diagnosis results, but in this case, the answers to the questionnaire contain one's own subjective opinion. In this case, even if one evaluates oneself as "extroverted," one may be objectively "introverted." In such cases, the personality diagnosis results may differ from the personality based on the objective impression, and the accuracy of matching may decrease.

[0064] On the other hand, when using the above-mentioned matching system S, a user who is looking for a partner may extract potential partners by using the above-mentioned matching system S. For example, a male user and a female user who are looking for a partner use the terminal device 10 to register for a matchmaking service provided by the service server 20. Then, each user measures vital data using vital sensor 15 while performing an action designated by matching server 40 (for example, watching the same content distribution program), and transmits the vital data to matching server 40.

[0065] The matching processing unit 408 of the matching server 40 extracts one male user from the multiple male users and performs a matching process for each of the extracted male users and one female user extracted from the multiple female users to determine the degree of matching. Then, the matching process is performed for each of the remaining female users in order to determine the degree of matching for each of them. From the degrees of matching thus obtained, multiple candidates are extracted from the female user with the highest degree of matching. Here, the next male user for whom the matching process has not been performed is extracted, and the matching process with each female user is similarly performed. In this way, the degree of matching can be determined for each combination of registered male users and registered female users, and partner candidates can be extracted.

[0066] Here, the matching degree is calculated for each combination of multiple male users and multiple female users, but it is also possible to specify the male users and female users to be subjected to the matching process and perform the matching process between the specified users.

[0067] According to this application example, it is possible to extract marriage partner candidates not from the viewpoint of conditions but from the viewpoint of emotions or personality, etc. Also, here, it is possible to narrow down the marriage partner candidates from the multiple candidates extracted by the matching process using conditions as necessary. In this case, even if many candidates with similar degrees of matching are found, it is possible to narrow down the candidates from among them taking conditions into consideration. Furthermore, this application example allows for the degree of matching to be obtained without including the user's own subjective factors, compared to personality assessments conducted through questionnaires, etc. Therefore, matching can be performed from a new perspective, based on emotions and values ​​corresponding to emotions. Furthermore, it is possible to narrow down the search by adding conditions as needed. In this way, marriage can be supported.

[0068] For example, when watching a movie, it may be said that users who obtained the same emotional data at the same scene in the movie have similar personalities. Also, even if different emotional data were obtained at the same scene in the movie, it is possible to estimate whether they have a relationship in which they can support each other, depending on the relationship with emotional data generated at other scenes.

[0069] Here, in the above-described matching system S, a case has been described in which an emotion estimation model that has learned the relationship between vital data and emotions is used, but another trained model may also be used. The learning device 30 may generate an estimation model of the matching degree by learning using, as training data, the relationship between emotion data estimated based on vital data acquired from the first user, emotion data estimated based on vital data acquired from the second user, and the matching degree between the first user and the second user.

[0070] More specifically, when applying the matching system S to a matchmaking service, learning can be performed as follows. For example, a married couple may be asked to watch the same movie, and the emotional data of each couple may be obtained based on their vital data, and the degree of compatibility (degree of matching) may be combined as training data to generate a trained model through learning by the learning device 30. In the case of a married couple, the compatibility used in the training data may be such that the longer the couple has been married, the greater the degree of compatibility (the higher the degree of matching). Furthermore, learning may be performed by further using as training data a combination of emotional data based on vital data about two people who actually met through a matchmaking service and the degree of compatibility between the two people (for example, the degree to which their relationship has progressed). Here, a value indicating a higher degree of compatibility (higher degree of matching) may be used the longer the couple has been together since actually meeting. Furthermore, vital data may be acquired from each couple who meet through a matchmaking service, and when the couple ends up getting married, the vital data may be used as training data to which a value indicating a high degree of compatibility is assigned, and a trained model may be generated. In this case, the shorter the period from the formation of the couple to marriage, the higher the degree of compatibility may be. In addition, a trained model may be generated by learning using at least two of the following: training data based on vital data obtained from a married couple; training data based on vital data obtained from couples who actually met and formed through a matchmaking service; and training data based on vital data obtained from two people who formed a couple through a matchmaking service and ended up getting married. In this way, by learning using vital data obtained from two people who are actually married, or from two people who have become a couple, etc., the relationship between the vital data that allows a couple to function as a couple, or the relationship between the vital data obtained from two people who have become a married couple, can be learned along with the degree of compatibility.Therefore, if there is such a vital data relationship, it can be learned that the compatibility is good, that is, whether the two people are compatible enough to function as a couple or to be able to live as a married couple.Therefore, by using such a trained model, it is possible to estimate the level of compatibility between two people whose compatibility you want to diagnose.

[0071] In this way, the relationship between the vital data of two people who are actually married, or the emotional data based on the vital data of two people who have formed and continued as a couple, and their actual compatibility is learned, and the matching processing unit 408 estimates the matching degree using this trained model (matching degree estimation model). As a result, the longer the two people being measured have actually been together, the better their compatibility is considered to be. Then, by learning the relationship between compatibility and the emotional data based on the vital data of the two people, the emotional data based on the vital data of the two people who are candidates for a couple can be input into the trained model, and the degree of compatibility can be obtained as the matching degree. Here, a trained model that takes into account the causal relationship between whether the relationship has actually continued for a long time is prepared and can be used. Here, the degree of matching may be expressed numerically, such as "compatibility: degree of suitability XX%" or "compatibility: degree of special existence XX%."

[0072] <Example of use: Performance by multiple people> In business (an organization within the same company (sales team, research team, etc.)), sports teams, etc., multiple people may form a single group (organization, team, etc.) and take action (sales activities, research, development, team sports). In such cases, it is believed that organizational performance is influenced not only by the skills of the individuals in the organization, but also by how well the team can strengthen its cohesion and how smoothly they work together. This type of performance is thought to be influenced not only by individual skills, but also by the compatibility between members. Therefore, the above-mentioned matching system S can be used to estimate the level of organizational performance. Here, the above-described matching process may be performed for each user belonging to an organization to determine the degree of matching, and the level of performance may be estimated based on the degree of matching. For example, the higher the degree of matching, the higher the estimation of performance.

[0073] Here, when the matching system S is used to evaluate the performance of an organization, it can be learned in the following manner. For example, each member of an already formed organization is asked to perform the same activity (light work, sports, etc.), vital data is acquired from each member while they are performing the activity, and emotional data is estimated based on the vital data. Then, using the combination of the level of performance (degree of matching) according to the organization's track record as training data, the learning device 30 learns and generates a trained model. For example, in a business team, the results of an evaluation of sales performance can be used as a measure of performance, and the better the sales performance, the higher the value indicating the higher the performance.Then, the relationship between each member's emotional data and the performance level based on the results of the sales activities carried out by that member is learned as training data. In addition, in a sports team, evaluation results based on game results can be used as a measure of performance, and the better the game results (for example, whether or not the team won against the opposing team, the total score, the overall ranking when participating in a tournament, etc.), the higher the value indicating the performance can be used. Then, the relationship between the emotional data of each member and the level of performance based on the results of sports played by that member is learned as training data.

[0074] In this way, the system learns using emotional data based on the vital data of an actual team and performance levels based on performance records, and then uses the trained model to estimate the degree of matching. As a result, the higher the performance level, the better the compatibility is deemed to be. Then, by learning the relationship between compatibility (performance level) and emotional data based on the vital data of the two people, the trained model can be input with emotional data based on the vital data of each member of the team, and the degree of compatibility can be obtained as the degree of matching. Here, by learning the relationship between the results of actual team activities and emotional data, a trained model that takes into account the causal relationship between emotional data and compatibility is prepared and can be used. This also makes it possible to consider the combination of team members based on the degree of match, from the perspective of whether, even if multiple highly skilled people come together, their individual skills will not be utilized due to differences in their personalities and policies, or whether, even if their individual skills are not so great, they can cooperate with each other and act in line with the same policies to achieve high performance as a team as a whole. The degree of matching may be expressed numerically, such as "compatibility: team performance ability XX%" or "compatibility: probability of achieving good results XX%."

[0075] According to the embodiment described above, the matching process is performed using vital data, so that devices that users can easily use can be used, the influence of individual user subjectivity is reduced, and matching between users can be performed with improved accuracy. Furthermore, the relationship between vital sensor measurement results and emotions is currently being analyzed, and instead of just estimating emotions from vital sensors, we use the estimated emotions to estimate compatibility. This makes it possible to estimate compatibility without being influenced by the user's subjective opinion, compared to conventional compatibility diagnosis based on questionnaire results.

[0076] The above-described matching server 40 may be a physical server or a cloud server provided by a cloud computing service. Furthermore, each function of matching server 40 may be provided in a single server device, or at least one of the functions may be implemented separately in a different device.

[0077] The matching server 40 in the above-described embodiment may be implemented by a computer. In this case, a program for implementing this function may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include devices that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or devices that store programs for a fixed period of time, such as volatile memory within the computer system serving as the server or client. The program may also be designed to implement a portion of the above-described functions, or may be capable of implementing the above-described functions in combination with a program already stored in the computer system, or may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0078] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention. [Explanation of symbols]

[0079] 10, 10a, 10b, 10n terminal equipment 15, 15a, 15b, 15n Vital Sensors 20 Service Server 30 Learning Device 40 Matching Server 401 Communications Department 402 Storage section 403 Content Provider 404 Acquisition Department 405 Scene division section 407 Emotion estimation part 408 Matching processing section 4021 Emotion estimation data storage unit 4022 Measurement target behavior memory unit NW Network S Matching System

Claims

1. an acquisition unit that acquires vital data, which is a measurement result of a vital sensor, for each of a plurality of target users; a matching processing unit that determines a matching degree between users based on the acquired vital data of a plurality of users; A matching system having:

2. The acquisition unit Acquire time-series vital data measured during the period when the behavior specified as the measurement target is being performed, The matching processing unit Among the vital data, vital data corresponding to a section to be matched is used to determine a matching degree between the users. The matching system according to claim 1 .

3. The time-series vital data can divide a period during which the behavior of the measurement subject is being performed into a plurality of different scenes, The matching processing unit Using vital data corresponding to a scene that is a target for matching among the scenes, a matching degree between the users is calculated. The matching system according to claim 2 .

4. an emotion estimation unit that estimates emotion data representing an emotion according to the vital data; The matching processing unit A degree of matching between users is calculated based on the emotion data estimated by the emotion estimation unit. The matching system according to any one of claims 1 to 3.

5. The emotion estimation unit Estimating emotion data for each of a plurality of indicators from the vital data; The matching processing unit Using emotion data for each of the plurality of indices obtained from the vital data, a degree of matching between the users is calculated. The matching system according to claim 4 .

6. The matching processing unit emotion data estimated based on vital data acquired from the first user; emotion data estimated based on vital data acquired from the second user; a matching degree between the first user and the second user; For a trained model that has learned the relationship between By inputting emotion data estimated from vital data for each of a plurality of users for whom the degree of matching is to be determined, the degree of matching between the plurality of users for whom the degree of matching is to be determined is determined. The matching system according to claim 4 .

7. emotion data estimated based on vital data acquired from the first user; emotion data estimated based on vital data acquired from the second user; a matching degree between the first user and the second user; A learning device that learns the relationship between

8. 1. A computer-implemented matching method comprising: acquiring vital data, which is a measurement result of a vital sensor, for each of a plurality of target users; Based on the acquired vital data of multiple users, the degree of matching between users is calculated. Matching method.

9. emotion data estimated based on vital data acquired from the first user; emotion data estimated based on vital data acquired from the second user; a matching degree between the first user and the second user; A learning method for studying relationships.

Images