JP2020038431A - Action recommendation device and action recommendation system - Google Patents

Abstract

To make an excellent action recommendation to a user and others by devising a way to process data.SOLUTION: An action recommendation device comprises: interview data transmission means to transmit interview data composed of a plurality of items to any user; answer reception means to receive an answer to the interview data for each of the items transmitted by the interview data transmission means; input value generation means to generate an input value obtained by changing a content of some of the items in the answer received by the answer reception means; output value acquisition means to input the input value generated by the input value generation means to an artificial intelligence device having an artificial neural network caused to learn a plurality of sets composed of mission data and the answer of a user having performed the mission data and acquires an output value in response to this input value; and mission data transmission means to transmit mission data based on the output value acquired by the output value acquisition means to any user.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a behavior recommendation apparatus and a behavior recommendation system, and more particularly, to a behavior recommendation apparatus and a behavior recommendation that recommend an action to be taken to a user by using an artificial neural network that has learned statistical data on a questionnaire to the user. About the system.

  Patent Literature 1 discloses that the present applicant transmits the n-th mission data including a plurality of missions to a plurality of user terminals, and transmits the n-th mission data transmitted from a user who has performed the n-th mission data. Means for receiving the user's mission achievement rate related to the n-th mission data and the n-th mission data, and improvement is found in the n-th interview data in consideration of the user's mission achievement rate for the n-th mission data. When there is an inquiry item to be transmitted, the user who transmitted the (n + 1) th mission data, which increased the load of at least one of the plurality of missions, to the user terminal, and performed the (n + 1) th mission data N + 1 interview data and the (n + 1) th missile respectively transmitted from Means for receiving the user's mission achievement rate pertaining to the n data, and considering the user's mission achievement rate pertaining to the (n + 1) th mission data, where the n + 1th interview data shows further improvement in the questionnaire. A means for determining that the one mission has contributed to the improvement of the questionnaire item. According to Patent Literature 1, it becomes possible to know with high accuracy what kind of user each mission is suitable for, and thereby, the incentive for the user to perform a mission that is beneficial to himself is also provided. Has the effect of working.

Japanese Patent No. 6152313

  According to the invention disclosed in Patent Literature 1, in order to provide a user with a finer and more fragmented, more useful mission, it is possible to transmit a wider variety of inquiry data to the user. In this case, the number of answers to the inquiry data transmitted from each user and the number of receptions of the mission achievement rate increase.

  Then, it is conceivable that the data processing time such as the mission achievement rate in the hypothesis verification device increases, and the cost increases accordingly. Further, the processing method of the mission achievement rate adopted in Patent Document 1 uses linear regression analysis. However, in this case, only about 10 items of the explanatory variables can be analyzed, and the mission achievement rate and the like can be analyzed. If the number of processes becomes unexpectedly large, it becomes almost impossible to perform the process in a realistic manner.

  Therefore, an object of the present invention is to prevent such a problem from occurring by devising a method of data processing.

The behavior recommendation device of the present invention,
An inquiry data transmission unit (for example, the communication unit 210 shown in FIG. 1) for transmitting inquiry data including a plurality of items to an arbitrary user;
Answer receiving means (for example, communication means 210 shown in FIG. 1) for receiving an answer of each item to the inquiry data transmitted by the inquiry data transmitting means;
Input value generating means (for example, calculating means 230 shown in FIG. 1) for generating input values obtained by changing some item contents of the answer received by the answer receiving means;
Inputting the input value generated by the input value generating means to an artificial intelligence device having an artificial neural network learned by a plurality of sets including mission data and answers from a user who performed the mission data. An output value acquiring means for acquiring an output value corresponding thereto (for example, a calculating means 230 shown in FIG. 1);
Mission data transmitting means (for example, communication means 210 shown in FIG. 1) for transmitting mission data based on the output value obtained by the output value obtaining means to the arbitrary user;
Is provided.

  According to the present invention, data based on an answer returned from a user is used as an input value, and an output value that cannot be realized by linear regression is calculated by using a learned artificial neural network, and mission data is calculated based on the output value. , It is possible to give an advanced mission to each user based on the task for each user.

  The mission data to be transmitted to the arbitrary user can be created based on the difficulty of performing the mission data used in learning the artificial neural network and an answer when the mission data is performed. .

  The artificial neural network can be an artificial neural network capable of performing temporal processing, including a recurrent neural network.

  The interview data includes the user's usual eating habits, for example, the frequency of eating out, and the frequency of eating lunches sold at convenience stores. In addition, the interview data includes some activities including the presence or absence of a usual headache, the presence or absence of a usual abdominal pain, ease of catching a cold, the presence or absence of a habit of moving a body, and the like.

In addition, the behavior recommendation system of the present invention,
The behavior recommendation device (for example, the administrator server 200 shown in FIG. 1),
An artificial intelligence device including the artificial neural network (for example, the AI device 30 shown in FIG. 1);
Is provided.

It is a schematic structure figure of an action recommendation system of an embodiment of the present invention. FIG. 2 is a diagram showing an example of inquiry data of each user registered in an inquiry data etc. DB 400 of FIG. 1. It is a figure showing an example of a result from a user who gave mission data. 3 is a flowchart illustrating an operation related to action recommendation among operations of the administrator server 200 illustrated in FIG. 1. 3 is a flowchart illustrating an operation related to action recommendation among operations of the administrator server 200 illustrated in FIG. 1. FIG. 2 is a diagram illustrating an example of a comparison table notified to a user created by a creating unit 260 in FIG. 1.

REFERENCE SIGNS LIST 100 User terminal 200 Administrator server 210 Communication means 220 Learning means 230 Calculation means 240 Registration means 250 Numericalization means 260 Creation means 300 Artificial intelligence (AI) device 400 Database for inquiry data (DB for inquiry data etc.)
500 administrator terminal

Embodiment of the Invention

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(Description of configuration)
FIG. 1 is a schematic configuration diagram of an action recommendation system according to an embodiment of the present invention. FIG. 1 shows a user terminal 100, an administrator server (behavior recommending device) 200, an artificial intelligence (Artificial Intelligence) device (hereinafter, referred to as an "AI device") 300, and a database for medical inquiry data, etc. (Hereinafter, referred to as “DB for medical inquiry data”) 400 and an administrator terminal 500.

  The user terminal 100 is a terminal operated by a user, mainly receives inquiry data transmitted from the administrator server 200 or mission data including a plurality of missions, and answers the inquiry data of the user or performs the mission. The latter response is transmitted to the administrator server 200 as appropriate.

  The user terminal 100 can be realized by a mobile phone, a smartphone, a personal computer, a video game, an IT device, an AV device, or the like, which can perform wireless communication or wired communication with the administrator server 200.

  The administrator server 200 is a server (action recommendation device) managed by the administrator of the action recommendation system of the present embodiment. The administrator server 200 includes a communication unit 210, a learning unit 220, a calculation unit 230, a registration unit 240, a digitization unit 250, and a creation unit 260, which will be described below. The function of each of these means may be realized by, for example, cooperation between a CPU and a memory.

  The communication unit 210 transmits various data such as medical inquiry data and mission data registered in the DB 400 such as medical inquiry data to the user terminal 100. In addition, the communication unit 210 receives various data such as an answer to the inquiry data transmitted from the user terminal 100.

  In addition, the interview data is composed of a plurality of items such as personal data such as sex, age, weight, etc., alcohol consumption, smoking amount, etc., which are to be answered by a subject or a patient at the time of a medical examination or initial consultation performed in a hospital or the like. . In addition, if possible, in addition to these, blood pressure, blood sugar level, furthermore, whether you are awake, whether you have a headache or abdominal pain, whether you have a chronic illness, how easy it is to catch a cold, whether you have a habit of moving your body, etc. You may ask for various information including. Furthermore, the user's usual eating habits, for example, the frequency of eating out and the frequency of eating a lunch box sold at a convenience store can be included, but the present invention is not necessarily limited to a medical consultation on health conditions.

  In addition, it is essential that the timing of transmitting the interview data is when the user uses the behavior recommendation system for the first time, and then, for example, every month or year from the date of use, or when there is a request from the user. It is conceivable to set any or all of the cases. In particular, if the notification is made on the same day every month or on the same month every year, it is possible to periodically give advice on health according to the result of the inquiry, which is preferable for the user.

  The learning means 220 uses, as learning data, a set of mission data (including the inquiry data) and the answer of the user who performed the mission (including the response corresponding to the inquiry data) stored in the DB 400 such as the inquiry data. This is used to make the artificial neural network of the AI device 300 learn, and for example, it is conceivable to input learning data through a programming API.

  Here, even if researchers, etc. are seeking monitors to collect the interview data and the answers to them, and the mission data and the effects such as the achievement rate, usually tens of thousands of monitors are required. Collecting numbers is very difficult.

  However, in the case of the action recommendation system of the present embodiment, if the user can operate the user terminal 100, mission data and effects such as the achievement rate can be collected. From a rate perspective, the total number of users is very large.

  In particular, as described later, the desired application software can be installed on the user terminal 100, and the achievement rate for the mission data can be transmitted by the administrator server 200 using the application software without explicitly transmitting the user. This is preferable.

  The calculating unit 230 sets, as an input value, a value obtained by changing some of the answers to the inquiry data transmitted to the user terminal 100 to the AI device 300 having the artificial neural network learned by the learning unit 220. Output value to be calculated.

  The registration means 240 is for registering various data of answers to the inquiry data / mission data transmitted from the user terminal 100 and received by the communication means 210 in the DB 400 such as the inquiry data. Then, learning of the artificial neural network of the AI device 300 can be continuously performed using the registered data.

  The digitizing means 250 digitizes various data transmitted to the user terminal 100 or the administrator terminal 500, for example, by digitizing the achievement rate of mission data by the user transmitted from the user terminal 100.

  The creating means 260 creates an information medium including the achievement rates of the mission data quantified by the quantifying means 250, such as the screen examples shown in FIGS. The screen created by the creating unit 260 is transmitted to the user terminal 100 through the communication unit 210 by a request from the user terminal 100 or voluntarily.

  The AI device 300 specifies a mission data to be transmitted to the user terminal 100 by utilizing a mathematical model such as an artificial neural network. In addition, as the artificial neural network, it is preferable to use a recurrent neural network that has strength in temporal processing, but is not limited to this.

  In particular, if an artificial neural network that has strength in temporal processing is used, it is possible to reduce the number of questions in the interview data or to provide higher-level recommended information by setting a unique task for each user.

  The inquiry data etc. DB 400 stores, as time passes, a set of mission data including inquiry data transmitted to the user terminal 100 and an answer of the user who performed the mission including an answer corresponding to the inquiry data transmitted from the user terminal 100. It is registered automatically.

  The administrator terminal 500 is a terminal operated by the administrator of the action recommendation system of the present embodiment. Although the administrator terminal 500 is not limited to this, for example, a desktop personal computer or the like can be used. By operating the administrator terminal 500, the administrator registers mission data to be transmitted to the user terminal 100 before the AI device 300 learns the artificial neural network. Specifically, for example, if there is a fact that it is better to eat an apple as a migraine countermeasure, an event of having a migraine and a mission to eat an apple are registered in association with each other.

  Note that, for example, FIG. 1 illustrates an example in which the AI device 300 and the inquiry data DB 400 are provided outside the administrator server 200. However, all or at least one of these is provided by the administrator server 200. It may be provided inside.

(Summary of action)
Next, the operation of the action recommendation system of the present embodiment will be outlined using a specific example. Note that a more detailed description of the operation of the action recommendation system of the present embodiment will be made in conjunction with the operation description described later.

First, for example, it is assumed that the inquiry data such as the following five items are initially transmitted to the user terminal 100 and the answer result is obtained.
A: Do you eat salads at least once a week?
Yes; eat No; do not eat B: Do you use stairs rather than elevators and escalators?
Yes; use No; do not use C: Do you drink less than 4 liquors per week?
Yes; less than 4 go No: more than 4 go D: At this time, isn't it possible to stay tired?
Yes; No No; Yes E: Are you awake?
Yes; good No; bad

  In the present embodiment, all the inquiry data can be answered with Yes / No, so that the burden on the user at the time of answering is reduced and the burden of data processing on the administrator server 200 is also reduced.

  It is assumed that the inquiry data composed of [ABCDE] is transmitted to the user α, and the answer result from the user α is “11001” when the answer is “1” for Yes and “0” for No. Then, since the amount of alcohol that the user α drinks in one week is large, it is conceivable that this may be a cause of keeping tired.

  Similarly, if the same inquiry data is transmitted to another user β and the answer result from another user β is [00101], the user β may have a low vitamin intake and / or lack of exercise. Is likely to be a cause of fatigue.

  However, the causal relationship between the answer results of the user α and the user β may be reversed. For example, the user β may not be able to secure enough time for breakfast due to poor sleeping and getting up, and thus the vitamin intake may be low.

  However, in the case of the user β, for example, encouraging the user β to increase the intake of vitamins and eliminate exercise deficiency may lead to an improvement in staying tired. Sex is useful because it is, and it would be worth trying.

  Therefore, in the present embodiment, a set of the inquiry data transmitted to the user α and the answer result [11001] of the user α, Similarly, the registration means 240 stores the set of the inquiry data transmitted to the user β and the answer result [00101] of the user β in the artificial intelligence network of the AI device 300 in the inquiry data etc. DB 400. Register.

  Further, as described later, the mission data to be recommended is transmitted to the user α and the user β based on the inquiry data, respectively, and the answers from the users α and β that have performed the mission data are also registered by the registration unit 240. The inquiry data etc. are registered in the DB 400. To give a typical example of the mission data and the answer, the user α is given the mission data of reducing the amount of alcohol to be drunk per week, and an answer that the user is less tired will be obtained.

  The number of users who use the action recommendation system of the present embodiment becomes very large. Therefore, if a very large number of inquiry data and the like are accumulated in the DB 400 such as the inquiry data, it is transferred to the artificial neural network of the AI device 300. Let them learn.

  Here, as for the example of the user β, if the artificial neural network of the AI device 300 is not sufficiently learned, data based on the interview data and the answer about the user β is input to the AI device 300. However, it is not possible to determine whether the low intake of vitamins and / or lack of exercise is likely to be a cause of tiredness. If the neural network has been sufficiently learned, inputting data based on the interview data and the answer to the user β into the AI device 300 indicates that the vitamin intake is low and / or exercise is insufficient. Thus, it is possible to determine whether the cause of the fatigue is high.

Specifically, the calculating means 230
Only the variable of item A regarding vitamin intake is “1” instead of “0” [101],
Only the variable of item B relating to lack of exercise is “1” instead of “0” [011],
[111] where both variables of item A and item B are “1” instead of “0”;
[001] in which the variables of both item A and item B are "1" instead of "0",
Are input values to the AI device 300, and respective output values corresponding to these are obtained.

For example, in the DB 400 for medical interview data,
When the number of registered answers in which [ABC] is [101] and [D] is [1] is 500,
-When [ABC] is [011] and [D] is [1], the number of registered answers is 500,
-When [ABC] is [111] and [D] is [1], the number of registered answers is 1000,
-When [ABC] is [001] and [D] is [1], the number of registered answers is 100,
Therefore, if [ABC] corresponding to [1000] having the highest accumulated number behaves as [111], it can be said that fatigue is statistically reduced with a high probability.

  In other words, generally speaking, when an answer to the inquiry data is obtained from a specific user, the calculating means 230 appropriately changes each item (variable) such as [ABC] in the above example. If these are input values to the learned AI device 300, output values corresponding to these can be obtained, and a mission based on the output values is given to the user.

  In this way, if the result of a case where the intake of vitamins is small, the exercise is insufficient, and it is statistically highly likely that the cause is not tired, it is possible to derive the interview data of a certain user γ. If the answer is that it is difficult to get tired, give user γ a mission to eat salad at least once a week and use stairs rather than elevators and escalators. When a mission such as shogi is given, it can be said that the user γ eliminates the cause of the fact that fatigue cannot be eliminated statistically.

  When the learned AI device 300 is used, a response is given to one user to take vitamins from vegetables, and another user is given vitamins to fruits by answering the inquiry data. Or the mission to do so.

  After assigning a mission to each user, the administrator server 200 acquires a response from the user who has performed the mission, and also accumulates a set of the mission and the answer in the DB 400 such as medical inquiry data.

  By the way, such a series of processes may increase the number of items of the interview data in order to provide a more useful mission that is detailed and subdivided. Then, it is not enough to derive what the preferred mission is from the four combinations of about 2 × 2 as in the above example, and the amount of calculation becomes enormous.

  Therefore, in the present embodiment, the artificial neural network of the AI device 300 learns the set of the mission including the interview data and the answer stored in the DB 400 such as the interview data, by the learning means 220, which is enormous. Even if it is a calculation amount, it is possible to derive what a mission is preferable for each user.

To summarize and generalize the above, as an outline of the action of the action recommendation system of the present embodiment, first,
(1) Send detailed and subdivided interview data to multiple (many) users,
(2) Assign missions that are considered favorable for each user,
(3) The artificial neural network of the AI device 300 is trained by using a set of the mission including the interview data and the answer when the mission is performed.

Next, after learning the artificial neural network of the AI device 300,
(4) Inputting a changed part (variable) of the answer transmitted from the specific user as an input value to the AI device 300 having the learned artificial neural network, and calculating a corresponding output value And
(5) selecting a mission based on the calculated output value to give a more useful advanced mission to the specific user;
It turns out that.

  Further, if a specific user performs a high-level mission given to him / her, and then the status of the specific user is improved / improved, the fact is stored in the DB 400 such as medical inquiry data. Will be improved.

  In addition, in the series of explanations up to this point, examples related to health such as whether to get tired or to wake up well were given, but this behavior recommendation system is not intended to improve health condition only It should be noted that by appropriately selecting the item contents of the interview data, it can be used in various fields.

  For example, but not by way of limitation, an advanced use menu for a particular athlete may be provided, for example, for an application. Specifically, if the objective variable related to the AI device 300 is replaced with a mileage or a sprint time, etc., if the guidance menu is given as mission data and the performance obtained by practicing it is used as an answer, It is possible to grasp whether the guidance menu should be selected. Similarly, by appropriately changing the objective variable, it is possible to specify the supplement to be taken by the athlete.

  Also, as a completely different example, if the objective variable is replaced with the sales of each employee, it is determined which job should be assigned to which employee to increase the sales of the entire company. Can also be used for Generally speaking, if the objective variable is relatively small despite the fact that the explanatory variable is relatively large, the action recommendation system can be suitably used.

(Description of application software)
Next, application software to be installed on the user terminal 100 will be described. Note that this application software does not necessarily have to be installed in the user terminal 100, and the presence or absence of the installation may be appropriately selected by the user.

  However, as a conclusion, if this application software is installed in the user terminal 100, there is an advantage that the operation burden is reduced for the user, and an arbitrary answer can also be eliminated for the action recommendation system. There are advantages.

  Here, for example, when the user terminal 100 is a smartphone, the smartphone typically has an imaging unit, and therefore, captures a face image of the user using the user terminal 100 or a meal of the user. be able to. When the user terminal 100 is a smartphone, the usage time, the talk time, the number of steps of the user carrying the user terminal 100, and the movement route (movement distance) using the GPS function are grasped. You can also.

  The application software transmits the captured image data of the face image, the image authentication result thereof, and the log data of the usage time of the user terminal 100 to the administrator server 200 as appropriate using the communication means of the smartphone. I do.

  If the image data of the face image transmitted from the user terminal 100 has not been subjected to image recognition processing, the administrator server 200 performs image processing on the image data, and determines whether or not the number of smiles has increased through matching processing or the like. It is determined whether the number of smiles per hour exceeds a predetermined threshold value, whether a salad is being eaten, and the like.

  Further, the administrator server 200 determines whether or not the daily number of steps has increased. As a result of these determinations, when it is confirmed that the number of steps of the user has increased, it can be determined that, in the case of the previous example, it is highly possible that the difficulty of getting tired is likely to be improved. .

  Further, when the user terminal 100 is a personal computer, if the personal computer is of a type having an imaging unit, use the personal computer. If there is a web camera connected to the personal computer, use the user. Similar to the case where the terminal 100 is a smartphone, the face image of the user and the like, as well as the posture, the time of the smile, the time at which the user will be talking, and the like may be transmitted to the administrator server 200.

  Furthermore, regarding the usage time of the user terminal 100, if a monitoring function of the task manager is installed in this application software, it is possible to grasp the usage frequency of the user's keyboard or mouse or the number of programs or files used. Therefore, the application software may transmit information indicating the usage time of the user terminal 100 and the like, such as image data of the face image, to the administrator server 200 as appropriate.

  As described above, by transmitting the information and the like obtained by using the application software almost in real time or every several minutes to several hours from the user terminal 100 to the administrator server 200, fatigue can be reduced by performing the mission early. It is possible to determine whether or not there is an improvement tendency in the health condition such as difficulty in falling out.

  If an attempt is made to obtain an answer from a user who has performed mission data without using application software, it is conceivable that the user will need to operate the user terminal 100 more frequently. This is preferable from the viewpoint of reducing the burden on the user.

  In addition, as an effect obtained by using the application software, an objective state determination based on image recognition or the like can be performed in the administrator server 200. It is possible to determine whether or not there is an improvement tendency in the health condition. Furthermore, since it is practically impossible for the user to recognize the number of smiles per unit time and transmit it to the administrator server 300 without using the application software, it is necessary to determine whether the number of smiles has increased. It is only necessary that the evaluation be transmitted from the user.

  However, even if the application software is not installed in the user terminal 100, if the inquiry data is transmitted to the user terminal 100 and the answer from the user is acquired as appropriate, the administrator server 200 can easily get tired. It can be determined whether or not is in an improving trend.

  Regardless of whether the application software is installed or not, whether the AI device 300 continuously learns the artificial neural network by successively updating the answer results to the inquiry data stored in the administrator server 200, An additional effect is also obtained that the mission data to be given to the server is preferably improved.

  Further, for example, it may be possible to find a causal relationship between a mission that is currently unknown and its effect. For example, it is not clear how to increase the number of smiles of the user per unit time, and as a result of performing many missions by a large number of users, the user smiles per unit time. If the number of smiles has increased, the mission may be statistically a factor in increasing the number of smiles.

  If so, as a secondary effect of using the action recommendation system of this embodiment, it is possible to investigate this causal relationship in the future, and furthermore, it may contribute to preventive medicine, etc. There is also a wider range of choices for future missions to be granted to other users.

  That is, as an example, it is assumed that, when an attempt is made to increase the number of smiles, for example, there are some missions that have been granted so far, but there are some circumstances that make it difficult to perform. On the other hand, there is a case where it is easy to perform a mission for which it has been found that the number of smiles can be increased by using the behavior recommendation system.

  In this way, the user can select the missions in the order in which the effects are considered to be high, or select the ones that have a high probability of achieving the mission in the past from those that are likely to have a certain effect. If you can do that, you will have more options for future missions.

(Description of Data Registered in Interview Data DB 400)
FIG. 2 and FIG. 3 are diagrams illustrating examples of the medical inquiry data of each user registered in the medical inquiry data etc. DB 400 of FIG.

  As shown in FIG. 2 as an example, the items of the interview data include “sex”, “age”, “weight”, “blood sugar level”, “blood pressure”, “good / bad” of “sleeping”, “good / bad” of “awakening”, and “good / bad” It shows items of "drinking amount" and "smoking amount". It should be noted that these items are merely examples, and the presence or absence of chronic illness including headache may be provided.

  In the answer column of the interview data, for example, five options can be prepared in order to simplify the input / selection of the interview data by the user. FIG. 2 and FIG. 3 show examples in which numerical values corresponding to this are registered. Specifically, for example, for gender, “1” corresponds to a male and “2” corresponds to a female, and for weight, “1” is too lean, “2” is lean, and “3” is lean. Usually, there are options such as "4" being slightly overweight and "5" being overweight.

  It should be noted that options that are generally considered good are set to have higher numerical values (in the above example, 5 points for “normal”, 0 points for “too thin”, 3 points for “slightly thin”, etc. .), For example, converting the interview data into a numerical value out of 100 is also effective because it leads to a numerical value target of the user.

  The arrows shown in FIG. 3 are created based on changes in the data example shown in FIG. Here, it is shown that the user who has a high “achievement rate” of the mission has improved the quality of “awakening”.

  Here, for convenience of explanation of the data registered in the interview data DB 400, the data example shown in FIG. 2 is shown as being overwritten to become the data example shown in FIG. Before completion of the learning of the neural network, there is a case where it is preferable not to overwrite the data because it is desired to increase the learning data.

(Explanation about learning data)
As described above, in the action recommendation system of the present embodiment, if the AI device 300 has not completed learning of the artificial neural network, the DB 400 asks the DB 400 for inquiry data and the like to accumulate learning data of the artificial neural network. A set of mission data, including data, and their answers will be accumulated.

The first mission to be given to user α for that is
(1) eat salad at least once a week,
(2) Use stairs rather than elevators and escalators,
(3) The number of cigarettes is limited to 5 per day,
Suppose there were three.

  Then, it is assumed that the user α properly performs the first mission, and then answers about the health condition. In this case, a set of the contents of each first mission and the answer thereto is stored in the DB 400 such as medical inquiry data.

Next, for example, a second mission having the following contents can be given to the user α. Here, an example is shown in which the second mission is obtained by partially changing the first mission, but may be completely different from the first mission.
(1) Eat salad at least twice a week,
(2) Use stairs rather than elevators and escalators,
(3) The number of cigarettes is limited to 5 per day,

  Actually, it is assumed that the second mission with the above contents is given to the user α, and then the user α completes the second mission, and then answers about the health condition. In this case, a set of the contents of each second mission and the answer thereto is stored in the DB 400 such as medical inquiry data.

Next, for example, a third mission having the following contents can be given to the user α. Although the third mission shows an example in which a part of the first and second missions is changed, the third mission may be completely different from the first and second missions.
(1) eat salad at least twice a week,
(2) Use stairs rather than elevators and escalators,
(3) Limit the number of cigarettes to "3" per day.

  Actually, it is assumed that the third mission with the above contents is given to the user α, and that the user α has also achieved the third mission, and then answered about the health condition. In this case, a set of the contents of each of the third missions and the answers thereto is stored in the DB 400 such as medical inquiry data.

  As described above, the content of the mission is changed as appropriate, and when a reply is made regarding the health condition thereafter, a set of the content of each mission and an answer to the mission is accumulated in the DB 400 such as medical inquiry data. . This is performed not only for the user α but also for other users.

  Then, a set of mission data including a large number of inquiry data and its answer will be accumulated for hundreds, thousands, or more users in the inquiry data DB 400. Using these as learning data, the artificial neural network of the AI device 300 learns.

  Then, for example, the inquiry data is transmitted to another user terminal 100, and when an answer corresponding to the inquiry data is returned from the user terminal 100, the calculating unit 230 sends the learned AI device 300 to the learned AI device 300. A preferred mission for obtaining some specific effect on the user terminal 100 based on this output value is to calculate a corresponding output value by using some of the answers to the interview data changed as input values. It is possible to select data and transmit it by the communication means 210.

  In other words, the behavior recommendation system can statistically grasp what kind of mission is performed and what kind of effect is obtained as a highly accurate result that is less affected by individual differences. This brings about an effect that mission data can be given to a user.

  Also, secondarily, for example, it is possible to give a more effective mission to a user, so that the user has a high probability of being confident that performing the given mission will be effective. As a result, an incentive to carry out the mission will work, and as a result, the action recommendation system can be operated in a good cycle in which the accuracy of the answer from the user fed back to the administrator server 200 increases. .

  Next, periodically or irregularly, for example, every week, the learning unit 220 uses the set of the mission data including the inquiry data accumulated in the DB 400 such as the inquiry data and the answer as learning data, Train the artificial neural network.

  Note that it is not essential that the number of missions is three, and it may be about five. However, if the number is too large, the mission achievement rate may decrease, so it is preferable to set the number to about three to five.

(Description of operation)
FIG. 4 and FIG. 5 are flowcharts showing operations related to action recommendation among the operations of the administrator server 200 shown in FIG. FIG. 4 shows a procedure before the completion of the initial setting among the actions related to the action recommendation. The processes shown in FIGS. 4 and 5 are performed for each user.

<Process before completion of initial setting>
First, the user operates the user terminal 100 to download the application software for the action recommendation system as described above, and installs the application software on the user terminal 100. Thereafter, if the user starts the application software, inputs communication information such as name, gender, age, and e-mail address and taps the transmission button, the transmission request of the inquiry data to the administrator server 200 is made. Is performed.

  In the administrator server 200, upon receiving the inquiry data transmission request from the user terminal 100 by the communication means 210 (step S11), the registration information 240 transmits the user communication information and the like transmitted together with the transmission request to the inquiry data etc. DB 400. And transmits the prepared inquiry data to the user terminal 100 that has requested the transmission (step S12).

  Thereafter, at the user terminal 100, the inquiry data is received. If the user inputs / selects his / her own inquiry data at the user terminal 100 and taps the send button, the user terminal 100 sends the inquiry data to the administrator server 200. Then, the inquiry data of the user is transmitted. Of course, the inquiry data may be transmitted to the user terminal 100 not by such a pull type but by a push type.

  In the administrator server 200, when the inquiry data is received by the communication unit 210 (step S13), the inquiry data is registered by the registration unit 240 in association with the communication information of the user already registered in the DB 400 such as the inquiry data. (Step S14).

  With the above processing, processing for initial setting such as member registration in the action recommendation system of the present embodiment is completed. The initial setting shown in FIG. 4 may be performed once for each user.

<Procedure after completion of initial setting>
FIG. 5 shows a procedure after the initialization up to step S14 in FIG. 4 is completed. First, the administrator server 200 selects mission data from the mission data registered in the DB 400 such as the inquiry data based on the answer to the received inquiry data (step S21).

  Here, as in the case where the operation of the action recommendation system of the present embodiment has just been started, if the AI device 400 has not yet sufficiently learned the artificial neural network, the answer to the received inquiry data is given. , The mission data registered by the administrator using the administrator terminal 500 is transmitted. In the example described above, if the user has a migraine, the mission data of eating an apple corresponding to the event of having a migraine is to be transmitted.

  On the other hand, if the AI device 400 has sufficiently performed the learning process on the artificial neural network, the calculating unit 230 sends the learned AI device 300 out of the results transmitted from the user who has performed the mission data. The corresponding output values are calculated by using some of them as input values.

This point is important, and will be described once again using other examples. For example,
・ Statistical data A that eating apples at least once a week relieves migraines,
・ Statistical data B that migraine is alleviated even if light exercise is performed two or three times a week,
・ Statistical data C that migraine hardly appears when eating apples at least once a week and performing light exercise several times a week;
Assume that there was.

Moreover,
Regarding Statistical Data A, the fact is that migraines are alleviated in 30% of people,
Regarding statistical data B, the fact is that migraines are alleviated in 40% of people,
Regarding Statistical Data C, it is in fact that migraines are alleviated in 95% of the people,
And

Then, from the user α who has completed the initial setting,
・ I do not eat much apples (variable 0),
・ Some lack of exercise (variable 0)
・ Has migraines,
Suppose there is an answer.

Here, if a human selects a mission, typically, for the user α, based on the statistical data C, eat an apple at least once a week and perform light exercise several times a week. It is expected that you will choose the mission. However, in this embodiment,
(1) A variable 0 indicating that an apple does not eat much is set to a variable 1 and input to the AI device 300 having the learned artificial neural network,
(2) The variable 0 indicating the lack of exercise is set to the variable 1 and input to the AI device 300 having the learned artificial neural network,
(3) Both variables 0 are set to variables 1 and input to the AI device 300 having the learned artificial neural network.
This is executed by the calculating means 230.

Then, since the AI device 300 is not a human, even if the effect itself of the statistical data A to C cannot be recognized, according to the calculating unit 230,
In the case of (1), an output value corresponding to the effect on 30% of the people is obtained,
In the case of (2), an output value corresponding to the effect on 40% of the people is obtained,
In the case of (3), an output value corresponding to the effect on 95% of the people is obtained.

  That is, the statistical data A to C are analysis results obtained by regression analysis, and the output values are analysis results obtained by deep learning of the learned AI device 300. In theory these would be equivalent.

  Therefore, based on these output values, the administrator server 200 can grasp that (2) is more preferable than (1) and (3) is more preferable than (2). Then, select the mission of eating the apple corresponding to (3) at least once a week and performing light exercise several times a week.

  The mission data is transmitted to the user terminal 100 by the communication unit 210 (Step S22).

  Thereafter, it is assumed that the user has performed a mission given from the administrator server 200. Typically, when the above-described application software is not installed in the user terminal 100, the user operates the user terminal 100 to receive a corresponding answer, for example, when new inquiry data is received. It can be input / selected and transmitted to the administrator server 200.

  On the other hand, typically, when the above-described application software is installed in the user terminal 100 and the application software is activated, the user's stepping or moving using the GPS function is appropriately performed on the user terminal 100. The distance and the like can be acquired and transmitted to the administrator server 200.

  Upon receiving the answer from the user, the administrator server 200 registers a set of the answer and the mission data related to the answer in the DB 400 such as medical inquiry data (step S23). Thereafter, it is determined whether or not the accumulated number of sets of the mission data and the answer in the medical inquiry data DB 400 is equal to or greater than a predetermined threshold (step S24). It should be noted that the threshold for the mission selection can be set by the administrator. The next mission data is transmitted to the user terminal 100 by the communication unit 210.

  As a result of the determination, if the set of the mission data and its answer in the medical inquiry data DB 400 is not equal to or greater than the predetermined threshold, the process proceeds to step S21 and the next mission data is selected. The n-th mission data and the (n + 1) -th secret data may have the same contents.

  On the other hand, as a result of the determination in step S24, if the accumulated number of sets of the mission data and the answers in the inquiry data etc. DB 400 is equal to or greater than the predetermined threshold, the learning means 220 stores the missions accumulated in the inquiry data etc. DB 400. Using the set of the data and the answer, a learning process for the AI device 300 is performed (step S25).

  By the above processing, a desired mission can be given to the user using the AI device 300 having the learned artificial neural network, and further, a set of the answer and the feedback given by the user, It is also possible to continuously learn artificial neural networks.

  Subsequently, the creating unit 260 performs various processes for displaying the information on the displays of the user terminal 100 and the administrator terminal 500 so that the user and the administrator can view the information described below, and creates display data.

  6 to 11 are diagrams showing examples of screens created by the creating means 260 shown in FIG. FIG. 6 shows an example of a TOP screen displayed after Taro who is a user logs in through a Web page to receive a service provided by the behavior recommendation system. Information necessary for login may be given to a user with a login ID and a password at the time of the initial setting described above.

  Here, as an example, a comment display area 610 showing a comment on Taro, an article display area 620 showing a special article for all users, and a food menu (mission data) recommended for Taro are displayed. A recommended food menu display area 630, a topic display area 640 in which various topics are displayed, a numerical display area 650 in which the achievement rate of Taro's mission data and the like are displayed, and Taro's health status is pseudo-converted to annual income. And a health status display area 660 in which a value and a rank are shown.

  The contents displayed in the comment display area 610, the recommended food menu display area 630, the numerical value display area 650, and the health state display area 660 are mainly created by the digitizing means 250. The contents displayed in the article display area 620 and the topics display area 640 are mainly input through the administrator terminal 500.

  FIG. 7A shows an example of a transition destination screen when the recommended food menu display area 630 in FIG. 6 is selected by clicking or the like. Here, the type of meal such as "Western food" or "Japanese food" is indicated, and "Western food" is selected by Taro. This shows a state where the menu is displayed. Each menu also displays a numerical value indicating how the pseudo annual income shown in the numerical value display area 650 and the health state display area 660 changes when they are eaten.

  These menus may be specific to Taro or may be common to all users. However, since the menu to be presented is mission data for Taro, it is not necessarily displayed to users other than Taro. More specifically, these menus will be presented to Taro and users with similar characteristics as Taro.

  The numbers attached to the menu are the basis for the pseudo annual income described above. This numerical value is calculated by the administrator server 200 by referring to the AI device 300 or the medical inquiry data DB 400 based on the achievement rate, that is, the degree of difficulty of the mission, and the effect of the mission. 500.

  Generally, it is conceivable that the higher the difficulty level and the higher the effect, the higher the numerical value. When utilizing such rules, even the most effective mission will not necessarily be the highest in the presented menu if the achievement rate is low.

  FIG. 7B shows an example of a pop-up screen when “Egg-rich omelet” in FIG. 7A is selected. The content of this example pop-up screen is the same as the content of FIG. 10A described later. However, the contents may be different.

  FIG. 8 shows an example of a transition destination screen when the health state display area 660 of FIG. 6 is selected by clicking or the like. Here, “$ 74,000,000” is shown in the upper left of the screen as a number corresponding to the pseudo annual income corresponding to the current Taro's health condition. At the center of the screen and at the lower left of the screen, as a basis of the quasi-annual income, it is additionally described whether the income is obtained by improving the diet or by improving the lifestyle.

  On the right side of the screen, the rationale based on the usage of the keyboard, mouse, and the like (Key and Mouse items, respectively), the rationale based on the number of open files, the number of programs used, and the like (File, Program, respectively) on the pseudo annual income In addition to the above, the grounds based on the work time, the number of smiles, the number of steps, the moving distance, and the like (the items of Time, Small, Walk, and GPS, respectively) are also shown. However, the display contents shown in FIG. 8 are examples, and information other than these may be shown as a basis or information different from these may be displayed.

  Here, it can be seen that the number of steps has decreased while the keyboard usage has increased. If the user of the desk work is busy and cannot secure sufficient exercise time, such a performance transition will occur. At this time, it is also a method to add a function for asking the user terminal 100 from the behavior recommending system, such as "Is the number of steps decreased, but were you busy?"

  In the case of steps, for example, the average number of steps, the maximum number of steps, and the minimum number of steps per unit period such as daily / weekly / monthly are acquired by measuring the number of steps daily, and the average It is determined that the degree of activity is high or low, and a high day is tabulated as “$ 1,000”, and a low day is tabulated as “$ -1,000”.

  In addition, typically, if the number of steps of the user's holiday such as Saturday and Sunday is also acquired, the number of steps of the holiday is generally considered to be smaller than the number of steps of the weekday. It would be a good idea to manage the number of steps and the number of steps on weekdays separately.

  In addition, on a day that exceeds the maximum number of steps, or when there is an increase of, for example, 20% or more with respect to the average number of steps, it can be estimated that a sudden action described later has occurred. In this case, for example, "$ + 2,000" can be set. On the other hand, when the number of steps is less than the minimum number of steps, or when the average number of steps is reduced by, for example, 20% or more, it can be estimated that there has been a visit to the hospital, which will be described later. In this case, for example, "$ -2,000" can be used.

  In addition, although the load is imposed on the user by forcing the user to perform the manual input, the user who carries the physical pedometer separately without using the pedometer function of the smartphone can use the pedometer through the user terminal 100. It is not excluded that the user manually inputs the number of steps measured in and transmits the number to the present behavior recommendation system.

  When the user terminal 100 is a video game, an IT device, an AV device, or the like, the number of times and frequency of use may be used for measuring the effect on mission performance, and these may be replaced with the information shown in FIG. Specifically, the user turns on the IT device or the like (further time until the device is turned off), logs in a predetermined log-in via the network (further time after the logout), and executes the IT device by the user. It is possible to indicate a numerical value corresponding to a pseudo annual income based on the number of times of action or the like.

  FIG. 9 shows an example of a transition destination screen when the numerical value display area 650 of FIG. 6 is selected by clicking or the like. It should be noted that, even when a location such as “see in chart” in FIG. 8 is selected by clicking or the like, the screen is changed to the screen example shown in FIG.

  Here, a screen example showing the usage transition of the keyboard, the mouse, and the like for one week shown on the right side of the screen in FIG. 8 is shown. Here, the usage transition and the like are shown in a line graph or a stacked bar graph according to the characteristics of the basis, but the type of the graph is not limited to these, and the user can appropriately select and select the display mode. It may be. Furthermore, instead of using a graph, the answer to the mission is represented by dots such as black and white, and the mission and the answer and the correlation with the answer are user It is one way to show it.

  FIG. 10A shows an example of a screen to which a transition is made when the “method” in the upper right of FIG. 6 is selected by clicking or the like. FIG. 10B shows an example of a screen to which a transition is made when the user selects “Details” in the article of FIG.

  The example of the screen in FIG. 10 (a) presents “Eat three eggs a day” to Taro as a mission, and when this mission is achieved, a number corresponding to the pseudo annual income described above. Shows how much is up. In addition, the effects and reasons expected by eating eggs, recommended articles for Taro, recipe information related to eggs, shop information, and the like are also displayed.

  FIG. 10 (b) shows a detailed article supervised by a dietitian. This example shows the meaning of the protein contained in eggs for the body, the goodness of menu such as omelet and steamed rice bowl, and the simple cooking method.

  FIG. 11A illustrates an example of a screen to which a transition is made when the user selects “My List” in the upper right of FIG. 6 by clicking or the like. FIG. 11A shows a my list display area 710 for transmitting a message when a mission is performed, and a result display area for displaying a response result based on the frequency of execution of each mission transmitted through the my list display area 710. 720, a breakdown display area 730 showing the breakdown of the entire habit-formed mission, and a mission display area 740 showing a list of missions given to Taro.

  The mission display area 740 displays date information on which a plurality of missions presented to Taro so far have respectively presented the missions, in addition to the mission “Eat three eggs a day” described with reference to FIG. In addition, information indicating the frequency of performing the mission with a star, information on the date on which the performance of the mission has become a habit, and information are also displayed.

  FIG. 11B shows an example of a screen to which a transition is made when a “detailed” portion in the my list display area 710 of FIG. 11A is selected by clicking or the like. Here, it is a guideline for determining the number of stars indicating the frequency of missions, an input method, and how the pseudo annual income changes depending on the determined number of stars. The explanation is shown.

When the screen examples shown in FIGS. 6 to 11 can be given to the user, the following effects can be expected.
-Since each user can be provided with a unique mission suitable for each user, an excellent effect for each user can be expected by performing each user's assigned mission.
-By providing missions to a large number of users and acquiring the performance results of each mission as feedback, it is possible to increase the effect measurement accuracy and to assign a proper unique mission suitable for the characteristics of the users.
-By presenting the transition of the actions familiar to the user (such as the frequency of using the keyboard) to the user after performing the mission, let the user easily understand the effect of the mission performance and continue the mission performance to the user. Incentives can be given.
-By devising the user interface (such as presentation of a pseudo annual income), it is possible to keep the user motivated to perform the mission.

  As described above, in the present embodiment, the present behavior recommendation system has been mainly described with an example of improving the health condition of the user. However, the effect of performing the mission is various, and in addition to the number of times the user smiles, The sleep time, the amount of meal, the transfer to the hospital, the sudden action, etc. can also be used for measuring the effect of performing the mission, so that, for example, the user may be able to select the items to be displayed on the right side of FIG. .

  It should be noted that, as a precautionary note, if the user is sick, moving to the hospital is a favorable action because it is an aggressive action for healing the user. Also, in order to properly judge the user's health condition on the behavior recommendation system side, it is good to know the onset history, hospital visit history, presence / absence of chronic illness, etc. Is also a law.

  This makes it possible, for example, to cope with a case where the user wants to view the disease history at an arbitrary timing. Presence or absence of movement to the hospital can be predicted based on the GPS function and the map application software provided in the smartphone or the like.

  Examples of the sudden action include, for example, going out on a trip or watching a sport or watching a movie that one does not usually go to. Such an action is considered to have a good effect on the mind and body of the user as a result, and thus it can be said that the action has a high degree of activity. For such an action, the numerical value corresponding to the simulated annual income should be increased. Therefore, if the administrator determines this numerical value, it may be increased. The presence or absence of a sudden action may be determined by input from the user or determination using a smartphone or the like, as in the case of going to a hospital.

  Further, it is preferable to determine whether or not a sudden action has occurred by using application software installed on a smartphone, because it is possible to acquire what and where the user is doing without bothering the user. However, for example, when both the tennis play and the tennis watching are selected as the items to be displayed on the right side of FIG. 8, in the determination using the application software installed on the smartphone, the user is required to enter the tennis court. , It is not possible to immediately grasp whether the player is playing tennis or watching tennis.

  For this reason, in the present embodiment, when the GPS function and the map application software specify that the user is on the tennis court, the identification is performed in consideration of the step count information obtained from the pedometer. I have. Specifically, when playing tennis, the possibility of putting a smartphone or the like in a pocket is not completely zero, but this possibility is not considered to be very high. If the total does not count the number of steps, it is determined that the player is playing tennis. Otherwise, it may be determined that the player is watching tennis. If it is difficult to determine any of these, the user may be selectively inquired.

  Similarly, when the user moves only within a predetermined range for a certain period of time or longer, such as when the user is browsing an exhibit such as a painting in an art museum, the GPS function and the map application software may be used. It is also possible to determine the level of activity by specifying the current location of.

  Furthermore, in recent years, there is a department store in which, for example, a futsal hall is provided on the roof to effectively use the roof of the building. Normally, it would be difficult to determine whether a user is visiting a department store for shopping or futsal purposes, but in the present embodiment, the GPS function and the map application software are used to determine the user's current location. The above determination is made by detecting not only the presence or absence of slight movement of the smartphone or the like in a short time, for example, one to two minutes.

  Specifically, for example, when the movement of the smartphone or the like is not confirmed by the GPS function of the smartphone or the like for 10 minutes, it is highly likely that the user has futsal with the smartphone or the like in a bag. Since it can be assumed, it is determined that futsal will be performed in this case.

  On the other hand, for example, when the movement of the smartphone or the like is confirmed by the GPS function of the smartphone or the like every one to two minutes, several tenant shops or the like are kept in a state where the user puts the smartphone or the like in the bag. It can be assumed that there is a high possibility that the user is determining the product while moving the. Therefore, in this case, it is determined that the user is shopping.

  Of course, although it is optional, it may not be possible to immediately determine which one is applicable. Therefore, in such a case, the user may be queried as in the case of tennis. The method of inquiry is unquestioned, and may be performed immediately when the determination cannot be made, or may be performed after a certain time has elapsed.

  In addition, inquiries to the user are considered to be highly likely to be playing futsal regularly, especially for users who regularly act in the same pattern, and therefore, when the request for transmission of the inquiry data is made, "Good o ○" ○ Do you play futsal on the roof of a department store? "

  If the user replies that "I play futsal for 1 hour from 15:00 every Saturday on the roof of the OO department store", then the GPS function and the map indicate that the user has entered the OO department store. When the confirmation is performed by the application software, it is possible to grasp the presence or absence and the type of the user's performance of the mission data without performing any extra judgment processing in this time zone.

It is a schematic structure figure of an action recommendation system of an embodiment of the present invention. FIG. 2 is a diagram showing an example of inquiry data of each user registered in an inquiry data etc. DB 400 of FIG. 1. It is a figure showing an example of a result from a user who gave mission data. 3 is a flowchart illustrating an operation related to action recommendation among operations of the administrator server 200 illustrated in FIG. 1. 3 is a flowchart illustrating an operation related to action recommendation among operations of the administrator server 200 illustrated in FIG. 1. FIG. 2 is a diagram illustrating an example of a comparison table notified to a user created by a creating unit 260 in FIG. 1. FIG. 7 is a diagram illustrating a screen example of a transition destination when the recommended food menu display area 630 in FIG. 6 is selected by clicking or the like. It is a figure which shows the example of a pop-up screen at the time of selecting "a lot of egg omelets" of Drawing 7 (a). FIG. 7 is a diagram illustrating an example of a transition destination screen when a health state display area 660 in FIG. 6 is selected by clicking or the like. FIG. 7 is a diagram illustrating an example of a transition destination screen when the numerical value display area 650 in FIG. 6 is selected by clicking or the like. FIG. 7 is a diagram illustrating an example of a screen to which a transition is made when a “method” at the upper right of FIG. 6 is selected by clicking or the like; It is a figure which shows the detailed article supervised by the dietitian. FIG. 7 is a diagram showing an example of a screen to which a transition is made when “My list” at the upper right of FIG. 6 is selected by clicking or the like. FIG. 12 is a diagram illustrating an example of a screen to which a transition is made when a “detailed” portion in the my list display area 710 in FIG. 11A is selected by clicking or the like.

Claims (5)

An inquiry data transmission means for transmitting inquiry data comprising a plurality of items to an arbitrary user,
Answer receiving means for receiving an answer for each item to the inquiry data transmitted by the inquiry data transmitting means,
Input value generating means for generating an input value in which some of the item contents of the answer received by the answer receiving means are changed,
Inputting the input value generated by the input value generating means to an artificial intelligence device having an artificial neural network learned by a plurality of sets including mission data and answers from a user who performed the mission data. Output value obtaining means for obtaining an output value corresponding thereto,
Mission data transmitting means for transmitting mission data based on the output value obtained by the output value obtaining means to the arbitrary user,
Behavior recommendation device equipped with.   The mission data to be transmitted to the arbitrary user is created based on a difficulty level of performing the mission data used in learning the artificial neural network and an answer when the mission data is performed. The behavior recommendation device according to 1.   The action recommendation device according to claim 1, wherein the artificial neural network is an artificial neural network capable of performing temporal processing, including a recurrent neural network.   The behavior recommendation device according to claim 1, wherein the inquiry data includes a user's usual lifestyle and health condition. An action recommendation device according to claim 1,
An AI device including the artificial neural network;
An action recommendation system with