WO2022004439A1

WO2022004439A1 - Learning assistance system, program, and learning assistance method

Info

Publication number: WO2022004439A1
Application number: PCT/JP2021/023267
Authority: WO
Inventors: 盛廣松本
Original assignee: 株式会社Airobo
Priority date: 2020-07-03
Filing date: 2021-06-18
Publication date: 2022-01-06
Also published as: JP7132579B2; JPWO2022004439A1

Abstract

A learning assistance system 1 has: an identification means 12 for identifying an on-line lecture; a schedule determination means 13 for determining a model schedule for attending the on-line lecture; a timing determination means 14 for determining a warning time with respect to the model schedule; and a warning means 15 for issuing, at the determined time, a warning to a user who is extracted from among users and whose attendance record shows lag behind the model schedule.

Description

Learning support systems, programs, and learning support methods

The present invention relates to a technique for supporting learning using an online course.

Technology related to learning progress management in online courses is known. For example, Patent Document 1 describes a technique of detecting a delay with respect to a predetermined learning plan and transmitting a message according to the delay. Patent Document 2 describes a technique for creating a course plan using the learning patterns of past successful applicants.

Japanese Unexamined Patent Publication No. 2006-350903 Japanese Unexamined Patent Publication No. 2004-259140

In the technology described in Patent Document 1, a warning was given from the viewpoint of whether or not the expected progress (that is, the expected date for completing the course) is in time for the predetermined end date. Further, in the technique described in Patent Document 2, a course plan was created using the learning patterns of past successful applicants.

On the other hand, in the learning support system, the present invention warns according to the attendance record at that time based on the comparison with the model schedule determined regardless of the specific attendance result among the users who have attended the course in the past. To provide the technology to do.

One aspect of the present disclosure includes a specific means for specifying an online course, a schedule determining means for determining a model schedule for taking the online course without using the attendance results of users who have taken the online course in the past, and the above-mentioned. A learning support system having a timing determining means for determining the timing of a warning for a model schedule and a warning means for giving a warning at the timing to a user extracted from users whose attendance record is behind the model schedule. I will provide a.

The warning means may give the warning to a user whose rate of change in the attendance record is lower than the standard.

The warning means may give the warning to a user whose course progress is relatively slow in a user group including a plurality of users.

The user group may be a group consisting of users who have taken the online course in the past or a group consisting of users who are actually learning according to the same model schedule.

The online course is divided into a plurality of chapters, the online course has a comprehension test after at least some chapters, and the attendance record may be specified according to the answer to the comprehension test. ..

The attendance record may be specified according to the time taken for the online course, or the progress rate or the enrollment rate of the online course.

The online course includes a plurality of subjects, and the attendance record may be specified according to the weight given to each of the plurality of subjects.

The schedule determination means may determine the model schedule using a machine learning model trained by giving the attendance record and the final result of the user who has taken the online course in the past as teacher data.

The timing determining means determines the model schedule using a machine learning model trained by giving the attendance record of a user who has taken the online course in the past, the timing of warning, and the effect as teacher data. May be good.

The warning means is a user who satisfies the conditions determined by using a machine learning model trained by giving the attendance record of the user who has taken the online course in the past, the target to which the warning has been given, and the effect as teacher data. It may be extracted as a warning target.

The warning means may determine the content of the warning using a machine learning model trained by giving the attendance record of the user who has taken the online course in the past, the content of the warning, and the effect as teacher data.

Another aspect of the present disclosure is a step of identifying an online course on a computer and a step of determining a model schedule for taking the online course without using the attendance results of users who have taken the online course in the past. , A program for executing a step of determining a warning timing for the model schedule and a step of giving a warning at the timing to a user extracted from users whose attendance record is behind the model schedule. I will provide a.

In yet another aspect of the present disclosure, the present invention provides a step by which a computer identifies an online course and a model schedule in which the computer takes the online course by a user who has taken the online course in the past. A step for determining without using the result, a step for the computer to determine the timing of warning to the model schedule, and a user for whom the computer is extracted from users whose attendance record is behind the model schedule. The present invention provides a learning support method having a step of giving a warning at the above timing.

According to the present invention, in the learning support system, it is possible to give a warning according to the attendance record at that time.

The figure which illustrates the functional structure of the learning support system 1 which concerns on one Embodiment. The figure which shows the specific system configuration example of the learning support system 1. The figure which illustrates the hardware configuration of the management server 10. A flowchart illustrating the operation of the learning support system 1. The figure which shows the composition example of an online course. The figure which exemplifies the attendance management database. The figure which exemplifies the change of the attendance record of a certain student. The figure which exemplifies the attendance record of a certain student group. The figure which exemplifies the display screen of the information about the attendance record. The figure which exemplifies the display screen of the information about the attendance record.

1 ... Learning support system, 2 ... Network, 10 ... Management server, 11 ... Storage means, 12 ... Specific means, 13 ... Schedule determination means, 14 ... Timing determination means, 15 ... Warning means, 16 ... Output means, 17 ... Provided Means, 20 ... client, 30 ... content server, 40 ... mail server, 50 ... analysis server, 91 ... object, 92 ... object, 93 ... object, 94 ... object, 95 ... object, 110 ... CPU, 120 ... memory, 130 … Storage, 140… Communication IF

1. 1. Configuration Figure 1 is a diagram illustrating the functional configuration of the learning support system 1 according to the embodiment. The learning support system 1 is a system that provides an online learning service. The online learning service is a service for distributing learning contents (that is, lectures) via a computer network. The learning support system 1 has a management server 10 and a client 20. The management server 10 is an information processing device that functions as a server in the online learning service. The client 20 is an information processing device that functions as a client in an online learning service.

The learning support system 1 has a storage means 11, a specific means 12, a schedule determination means 13, a timing determination means 14, a warning means 15, an output means 16, and a provision means 17. The storage means 11 stores data and programs. The data stored in the storage means 11 includes a course management database. The attendance management database is a database in which data for managing attendance status is recorded. The attendance management database records a user's profile (ie, attributes) as well as information about the courses the user has taken and is taking.

The identification means 12 specifies the online course to be taken. The online course refers to learning content distributed via a computer network and learning using a computer device. The schedule determination means 13 determines a model schedule for taking an online course. The model schedule is a schedule that serves as a reference for achieving the learning goals of online courses. The timing determination means 14 determines the timing of the warning for the model schedule. The warning means 15 warns the students extracted from the students whose attendance record is behind the model schedule at a determined timing. The output means 16 provides the student and / or other users with information according to the attendance status. The providing means 17 provides learning contents to the students.

FIG. 2 is a diagram showing a specific system configuration example of the learning support system 1. In this example, the learning support system 1 has a management server 10, a content server 30, a mail server 40, and an analysis server 50 as devices for providing services. In this example, the user of the learning support system 1 is divided into a learner and an administrator. A student is a person who takes a course, for example, a student at a school, a cram school, or a preparatory school, or an office worker. The manager is a person who manages the students, for example, a teacher at a school, a cram school, or a preparatory school, or a person in charge of the employee education department of a company employee or a boss. The client 20 includes a device used by the administrator and a device used by the student.

The content server 30 stores learning content (for example, a moving image) of an online course. The content server 30 distributes these learning contents to the students. The mail server 40 sends an email to the students and / or the administrator. The analysis server 50 analyzes the model schedule and the warning timing. The student accesses the management server 10 from the client 20 via a network 2 such as the Internet. Similarly, the administrator also accesses the management server 10 from the client 20.

FIG. 3 is a diagram illustrating the hardware configuration of the management server 10. The management server 10 is a computer device having a CPU (Central Processing Unit) 110, a memory 120, a storage 130, and a communication IF (Interface) 140. The CPU 110 is a device that executes processing according to a program. The memory 120 is a storage device that functions as a workspace when the CPU 110 executes processing, and includes, for example, a RAM (RandomAccessMemory) and a ROM (ReadOnlyMemory). The storage 130 is a storage device for storing data and programs, and includes, for example, an SSD (Solid State Drive) or an HDD (Hard Disk Drive). The communication IF 140 communicates with other computer devices according to a predetermined communication standard (for example, Ethernet (registered trademark)).

In this example, the program stored in the storage 130 includes a program for making the computer device function as the server of the learning support system 1 (hereinafter referred to as "server program"). When the CPU 110 executes the server program, the function shown in FIG. 1 is implemented in the computer device.

At least one of the memory 120 and the storage 130 is an example of the storage means 11 in a state where the CPU 110 is executing the server program. The CPU 110 is an example of a storage means 11, a specifying means 12, a schedule determining means 13, a timing determining means 14, a warning means 15, an output means 16, and a providing means 17.

Although detailed description is omitted, the client 20 is a computer device having a CPU, memory, storage, display device, input device, and communication device, for example, a personal computer, a smartphone, or a tablet terminal. The client 20 accesses the management server 10 or the like via a program dedicated to the online learning service or a general-purpose web browser program.

2. 2. Operation FIG. 4 is a flowchart illustrating the operation of the learning support system 1. In the following description, functional components such as the specific means 12 may be described as the main body of the process, in which the CPU 110 executing the server program executes the process in cooperation with other hardware elements. Means to do.

In step S1, the identification means 12 specifies the online course to be taken by the user. The online course to be taken by a user is automatically determined by the user himself / herself, the user's manager, or the learning support system 1. For example, if the student is a new employee of a company, the online course that the student takes is determined (or designated) by an employee of the personnel department. At this time, the profile of the user who is a student is registered in the learning support system 1 in advance or after this. A user's profile includes, for example, the user's age, gender, educational background, past exam grades, performance appraisal, and the like.

FIG. 5 is a diagram showing a configuration example of an online course. Here, a training program for new employees of financial institutions will be described as an example. Online courses consist of multiple subjects. In the example shown in the figure, the online course consists of four subjects: business etiquette, harassment eradication, sales representative qualification examination preparation, and FP (financial planning) skill test preparation. Each subject is divided into multiple chapters (ie, includes multiple chapters). Tests are provided after at least some of these chapters. In this example, the online course includes four types of tests: comprehension test, mid-term test, ability test, and mock test. The comprehension test is provided at the end of a predetermined unit (for example, each chapter) and is a test for confirming the comprehension of the unit. The mid-term test is provided after several chapters (and has subsequent chapters), and is a test to confirm the comprehension of these multiple chapters. In each subject, the part from Chapter 1 to the final chapter (including the mid-term test) is called the "main course". The ability test is provided after all chapters and is a test to confirm the comprehension of what is learned in the subject. The mock test is, for example, a test provided in the subject of qualification test preparation, and is a test for confirming the comprehension level in a question format imitating the qualification test. At least some of these tests may be omitted. At least a part of these four types of tests may be executed at any time regardless of the progress of the main part of the course. Also, if the student's score does not reach the reference point in at least some of these four tests, the text or question for that chapter may be displayed or the video for that chapter may be played again. good. At the time of adding these, the analysis server 50 processes by machine learning from the attendance results and results of a plurality of students who have already taken the course in the past.

Refer to Fig. 1 again. The providing means 17 provides an online course to the students. For each chapter, the providing means 17 records the date and time when the end criteria (for example, a passing score is obtained in the comprehension test), the date and time when the test such as the comprehension test is performed, and the score thereof in the attendance management database. Record for each student at. In this way, the history of attendance (when and which chapter of which subject was completed and the scores of various tests) and the history of warnings are recorded in the attendance management database. In addition, the attendance management database contains the attendance record of each student. The attendance record is a parameter that quantifies the status of attendance at that time. In one example, the main part of the course is divided into multiple categories in each subject. Weights are set for the main part, the ability test, and the mock test, and the weight is multiplied by the progress rate of each (for example, the pass in the comprehension test, the score above the reference point in the ability test, and the rate of completing the test in the mock test). A value (for example, a weighted average value) obtained by integrating the obtained values (hereinafter referred to as “study rate”) for a plurality of subjects is used as this parameter.

The progress rate of the main part indicates the ratio of the number of chapters that have been taken (for example, the chapter that got a passing score in the comprehension test) to the number of all chapters that should be taken. For example, if 25 chapters have been taken in a course consisting of 50 chapters in all 4 subjects, the progress rate is 50%. Alternatively, the progress rate may be the ratio of the total playback time of the chapters taken to the total playback time of the videos included in the course. Further, the progress rate may indicate the ratio of the number of chapters taken to the number of chapters to be taken up to that point in the model schedule. For example, if the model schedule indicates that the student has completed Chapter 5 on May 31, and the student has completed Chapter 4, the progress rate of the student is 80%. Regarding the ability test, for example, the progress rate is set to 100% by obtaining a score exceeding a preset reference point. For the mock test, for example, the progress rate may be 100% by completing the examination, or the score division may be set in advance and the weighted progress rate may be calculated for each score division. When taking multiple courses, if the integrated course rate (that is, the course rate obtained by multiplying the course rate of each course by the weight) exceeds the preset standard value, it is judged that all of those courses have been completed.

In this example, each course or subject has a deadline (or end schedule), and if you do not complete the course or the course by this deadline, you will not be able to take the course. Even if I take the course, the record is not reflected in the progress. In addition, in each course or subject, the criteria for determining that the course or the subject has been completed (or passed) can be individually set by the provider of the course or subject. In one example, the course or course provider sets a reference value for the enrollment rate. Learning support system 1 is. This enrollment rate is calculated for each student, and if the enrollment rate exceeds the standard value, it is judged that the enrollment has been completed (or passed).

Furthermore, the result of taking the course is recorded in the course management database. The results of taking the course include, for example, the pass / fail of the corresponding qualification test, or the score of the final test. In this way, the course history, the history of warnings, and the result of the course are recorded in the course management database.

FIG. 6 is a diagram illustrating an attendance management database. In this example, the attendance management database records the user's profile, course name, and attendance record of each subject. Regarding the user profile, only the user name is shown in the figure, but various other attributes may be recorded in the attendance management database. The attendance record for each subject includes the course and integrated enrollment rate, as well as the final study date, comprehension test score, and completion flag (ie, pass / fail) for each chapter.

Refer to FIG. 4 again. In step S2, the schedule determination means 13 determines a model schedule for attendance for the identified online course. The model schedule is determined, for example, before starting the online course. In one example, the schedule determination means 13 determines the model schedule according to a predetermined rule. For example, the storage means 11 stores a rule base in which one or more rules for determining a model schedule are recorded. The schedule determination means 13 determines the model schedule with reference to this rule base. In one example, this rule determines the model schedule based on the number of chapters contained in the course and the number of days until the scheduled end date of the course. That is, this rule does not use the attendance results of users who have taken this online course in the past (more specifically, the attendance record of users who have achieved a specific attendance result among the users who have taken this online course in the past. It determines the model schedule (without using it). The attendance result referred to here includes at least one of the final result of the course (for example, passing or failing) and the final goal of the course (for example, passing in a certain qualification examination). A particular attendance result is, for example, the result of passing the course or the corresponding qualification test. In addition, "without using the attendance results of users who have taken this online course in the past" means the attendance record of all users who have taken this online course in the past (not only the users who achieved a specific attendance result). In more detail, it does not exclude referring to the time change of the attendance record, that is, the actual attendance schedule).

More specifically, for example, if this course consists of 50 chapters in 4 subjects and 25 days until the scheduled end date of the course, the schedule determination means 13 determines the model schedule so that 2 chapters are taken per day. In the rules, priorities are stipulated for the subjects included in the course (for example, in the order of sales representative qualification test measures> FP skill test measures> business etiquette> harassment eradication), and the schedule determination means 13 follows these priorities. Determine the model schedule. For example, Chapter 1 of the sales representative qualification examination measures and Chapter 1 of the FP skill test measures (above 1st day), Chapter 1 of business etiquette and Chapter 1 of harassment eradication (above 2nd day), sales representative qualification The model schedule of Chapter 2 of the test preparation and Chapter 2 of the FP skill test preparation (above the third day), etc. will be decided.

Alternatively, the schedule determination means 13 may determine the model schedule according to the instruction of the user (specifically, the administrator). That is, the administrator may create a model schedule and input the model schedule created by himself / herself into the learning support system 1.

In one example, the model schedule is determined in units of user groups. That is, a plurality of students belonging to the same user group share one model schedule. The schedule determination means 13 stores information indicating the model schedule in the storage means 11 together with the identification information of the user group. In another example, the model schedule may be determined on a per-user basis. In this case, the schedule determination means 13 stores the information indicating the model schedule in the storage means 11 together with the identification information of the user.

In step S3, the timing determination means 14 determines the warning timing in the determined model schedule. The timing of the warning is determined, for example, before the start of the online course. In one example, the timing determining means 14 determines the timing of the warning according to a predetermined rule. In one example, this rule is based on the structure of the online course. Specifically, the main part of the course illustrated in FIG. 5 is divided into a plurality of (for example, four at the maximum) categories. This rule defines the date when each division ends in the model schedule as the warning timing. In another example, this rule is associated with a calendar and provides for warnings at predetermined time intervals. Specifically, it is stipulated that warnings be given daily, weekly, monthly, or a combination thereof. In another example, the rule for determining the timing of the warning is associated with the rule for determining the model schedule in the rule base. The timing determining means 14 determines the timing of the warning with reference to the rule used when determining the model schedule and the corresponding rule. In one example, this rule defines the date of the mid-term test for each subject in the model schedule as the timing of the warning. In yet another example, the timing of the warning may be specified by a user (specifically, for example, an administrator).

When the warning timing comes, the warning means 15 gives a warning to the students extracted as the target (step S4). In one example, the warning means 15 extracts a target student according to a predetermined rule (or determination condition). This rule is associated with, for example, a rule for determining a model schedule in a rule base. The warning means 15 refers to the rule used when determining the model schedule and the corresponding rule, and extracts the students to be warned. In one example, this rule is a rule (or condition) for further narrowing down the students from among the students whose attendance record is behind the model schedule. In one example, the warning means 15 extracts a student whose change rate of the attendance record is lower than the standard as a warning target. In this example, the attendance record of each student is stored in the attendance management database. The warning means 15 refers to the attendance management database and narrows down the target attendees.

FIG. 7 is a diagram illustrating changes in the attendance record of a certain student. In this figure, the horizontal axis shows the date and the vertical axis shows the attendance record. In this figure, the broken line shows the model schedule, and the solid line shows the attendance record. In this example, after the xth month and the xth day, the attendance record began to lag behind the model schedule, and on the y month and y day, the rate of change in the attendance record (that is, the slope of the solid line in the figure) fell below the standard. If the judgment of step S4 is made every day, this student is extracted as a warning target on y month y day. In another example, this student may be extracted as a warning target when the attendance record begins to lag behind the model schedule (the day following the x month x day). Alternatively, when the degree to which the attendance record is behind the model schedule (for example, the difference between the actual enrollment rate and the ideal enrollment rate in the model schedule) exceeds the standard, this student is extracted as a warning target. You may.

In another example, the warning means 15 extracts a student whose progress is relatively slow in a predetermined student group (an example of a user group) as a warning target.

FIG. 8 is a diagram illustrating the attendance record of a certain student group. The student group is, for example, a group of students determined by the administrator, and in one example, it is an employee who joined the company at the same time (an example of a group consisting of users who are actually learning according to the same model schedule). In this figure, the horizontal axis shows the attendance record and the vertical axis shows the number of students. The attendance record shows the progress rate. The warning means 15 extracts the students (hatched portion in the figure) whose progress is relatively slow (for example, the lower 10%) in this student group as the target of the warning.

The attendance record is not limited to the examples shown in FIGS. 7 and 8, and any attendance record may be used as long as the attendance status at that time is quantified. For example, the attendance record may be the average score of the comprehension test of each chapter. Alternatively, the attendance record may be a parameter that comprehensively considers a plurality of factors such as the number of chapters that have passed the comprehension test, the viewing time of the video, and the score of the comprehension test. In addition, a weight is set for each subject, and the value obtained by multiplying the number of chapters (or other parameters or a combination thereof) exceeding the reference point in the comprehension test by the weight may be used as the attendance record. This weight is set in advance for each course, for example.

The warning means 15 gives a warning to the students extracted as the target. This warning includes messages that promote learning, such as "Attendance is behind the model schedule. Let's do our best" and "Attendance is behind other students. Let's do our best". ..

The warning means 15 may give a warning to the administrator of the student in addition to or instead of the student. This warning is, for example, "Mr. XX's attendance record is behind the model schedule. Let's encourage her to do her best" or "Mr. XX's attendance record is considerably behind in the group who joined the company at the same time." Etc., Includes a message reporting the student's attendance status. In one example, the warning means 15 gives a warning using an e-mail message. The transmission of the e-mail is performed via the mail server 40. In another example, the warning means 15 gives a warning using SNS. The warning means 15 records information indicating the timing of the warning and the content of the warning in the attendance management database.

The output means 16 outputs information regarding the student's attendance record to the student or the administrator. The information about the attendance record includes the progress of the attendance (that is, the information that identifies the chapter that has been attended) and the test result, as well as the information showing the analysis result of the attendance record. The analysis result of the attendance record includes, for example, information for identifying the weakness or weakness or subordinated subject (subject with a score below the passing average) of the student, and information indicating the possibility that the student will pass the corresponding examination. include.

FIG. 9 is a diagram illustrating a display screen of information regarding the attendance record. Here, an example is a screen of My Page in which a student manages his / her own attendance record. In this example, the My Page screen has a plurality of display objects of object 91, object 92, object 93, object 94, and object 95. The object 91 is an object that visually shows the attendance record of the student. In this example, it is shown in the graph exemplified in FIG. This graph is displayed by switching the attendance record for the entire course or for each subject by tabs.

The object 92 is an object that displays the goal and the possibility of passing the course. The goals of the course are predetermined for each course. The passability is determined by the learning support system 1 using the machine learning technique described in the modified example described later. Alternatively, passability may be determined based on the relative grades of all students taking the course.

Object 93 is an object that analyzes subjects that it is not good at. When the object 93 is selected, the output means 16 outputs the analysis result of the subject (or a specific field) that the student is not good at. The results of this analysis may include links to additional learning content to overcome weak subjects (or areas of weakness).

Object 94 is an object showing the result of group analysis. When the object 94 is selected, the output means 16 outputs the result of the group analysis. Group analysis is information that shows the relative grades of the students in the group of students who are taking the same course. The results of the group analysis include statistical information on attendance performance in this student group (eg, the average number of learned chapters at that time, the average score of the comprehension test, etc.).

Object 95 is an object that displays a message from the administrator. When the object 95 is selected, the output means 16 displays a history of messages from the administrator. Screens that display messages from administrators include UI objects for entering replies to these messages.

FIG. 10 is a diagram illustrating a display screen of information regarding the attendance record. Here, an example is a screen in which the administrator manages the attendance record of the student group. This screen has a plurality of display objects of object 81, object 82, object 83, object 84, and object 85. The object 81 is an object that visually shows the result of the group analysis. In this example, the result of the ability test of a certain subject in a certain course is shown by a three-dimensional bar graph. In this graph, the x-axis shows the progress rate of a certain subject, the y-axis shows the score in the ability test, and the z-axis shows the number of people. The manager can know the degree of understanding of the students and the progress of the courses in the entire student group by, for example, looking at the screen of the analysis result as shown in the figure.

The analysis result displayed on the object 81 is not limited to the example of FIG. For example, the x-axis of a three-dimensional bar graph may indicate the weighted progress rate. The weighted progress rate is a value obtained by multiplying the progress rate of a plurality of subjects constituting the course by a weight. Alternatively, the y-axis of the three-dimensional bar graph may indicate the pass / fail result of the qualification test (or the score if the score is published).

The object 82 is an object that displays the goals of the student group and their achievability. The goals of the student group are predetermined for each group. The achievability of the goal is determined by the learning support system 1 using the machine learning technique described in the modified example described later.

Object 83 is an object that analyzes subjects that are not good at it. When the object 83 is selected, the output means 16 outputs the analysis result of the subject (or a specific field) that the student group is not good at.

Object 84 is an object showing the result of group analysis. When the object 94 is selected, the output means 16 outputs the result of the group analysis. Group analysis is information that shows the distribution of student attendance records and test grades belonging to the student group. The entire student may be divided into multiple subgroups and each subgroup may be analyzed. In this case, the results of attendance and test results may be compared with other subgroups. For example, the subgroup to which the student belongs may be changed according to the progress and grades of the student.

Object 85 is an object that displays a message from the student. When the object 85 is selected, the output means 16 displays the history of messages from the students. Screens that display messages from students include UI objects for entering replies to these messages. The administrator can exchange messages with the student one-on-one or one-on-one via this UI object.

By referring to this information, the administrator can take measures to promote follow-up and other attendance to students who have delayed attendance. This makes it possible to improve, for example, the passing rate of the qualification test. The output means 16 outputs a management screen in response to a request from a student or an administrator.

Note that the management screens of FIGS. 9 and 10 are merely examples, and the information output by the output means 16 is not limited to this. At least a part of objects 91 to 95 may be omitted, or objects other than these may be added.

3. 3. Modifications The present invention is not limited to the above-described embodiment, and various modifications can be carried out. Hereinafter, some modification examples will be described. Two or more of the items described in the following modification may be applied in combination.

3-1. Modification 1 (schedule determination means)
The method by which the schedule determining means 13 determines the model schedule is not limited to that exemplified in the embodiment. The model schedule is determined using, for example, one of the following methods.
(1) The model schedule is determined using the rule base (example described in the embodiment).
(2) The learning support system accepts the input of the model schedule created by the user (for example, the administrator), and decides to use the accepted model schedule. That is, the user manually determines the model schedule (another example described in the embodiment).
(3) Determine the model schedule using machine learning.
(4) Determine the model schedule using the attendance records of multiple users belonging to the user group.

Explain an example of determining a model schedule by machine learning. This machine learning model has an input layer, an intermediate layer, and an output layer. The learning support system 1 provides a course to this machine learning model by using the data of the course schedule recorded as the course record of a plurality of students who have taken the course in the past and the profile of the student as an input layer. The data of the result (or the final result) of the course is given to the output layer as teacher data to perform machine learning. As an example, the attendance schedule data is data in which "1" is entered on the day when the chapter is taken and "0" is entered in the matrix where the chapters of the subjects included in the course are arranged in rows. be. The final result data is data showing the pass / fail of the qualification test, which is the goal of the course. This machine learning process is performed by, for example, the analysis server 50.

The schedule determination means 13 determines the model schedule using the trained model thus obtained. For example, the schedule determination means 13 generates a plurality of candidates for the model schedule according to a predetermined rule. The schedule determination means 13 sequentially inputs these plurality of candidates into the trained model together with the profile of the target student. For this input, the trained model outputs the expected result, for example, the passability of the target qualification test. The schedule determination means 13 selects a plurality of candidates whose pass probability exceeds the reference value and determines them as a model schedule to be presented to the students. The reference value is, for example, predetermined (80%, etc.). Alternatively, this reference value may be determined by the manager or the student.

By determining the model schedule using machine learning in this way, it is possible to present an optimized model schedule based on the past achievements of the students.

An example of determining a model schedule using the attendance records of multiple users belonging to a user group will be described. In this example, the model schedule is determined using the attendance record of the course currently being taken (not yet completed) in a predetermined user group (for example, a user group of employees who joined the company at the same time). That is, in this example, the model schedule itself is not fixed before the start of the course (it is decided to determine the model schedule using the course record of the users belonging to this user group).

In one example, the model schedule is determined using the statistical representative value of the attendance record of the users belonging to the user group. For example, when the attendance record in the user group is the distribution as shown in FIG. 8, the statistical representative value of this distribution is used as the model schedule. As the statistical representative value, for example, an average value, a mode value, or a high-order x% value (for example, a high-order 30% value) is used. In a more detailed example, if the average value of the attendance record of the user group at that time is 25 out of 50 chapters in all 4 subjects, 25 chapters is the model schedule.

3-2. Modification 2 (timing determination means)
The method by which the timing determining means 14 determines the timing of the warning is not limited to that exemplified in the embodiment. The timing of the warning is, for example, at least one of the following.
(1) Timing determined based on the structure of the online course. For example, the day when the division determined in the model schedule ends (example described in the embodiment).
(2) Timing determined based on the calendar (daily, weekly, monthly, etc. Another example described in the embodiment).
(3) Timing determined based on the rule base and model schedule. For example, the day on which the midterm test is performed in the model schedule (another example described in the embodiment).
(4) Timing specified by the user (another example described in the embodiment).
(5) Determination of timing by machine learning.

An example of determining the timing by machine learning will be explained. The learning support system 1 gives the machine learning model the data of the course schedule recorded as the course record of a plurality of students who have taken the course in the past, the profile of the student, and the timing at which the warning is given. Is given to the input layer, and data indicating the presence or absence of the effect after the warning is given to the output layer as teacher data to perform machine learning. The data indicating the presence or absence of the effect indicates, for example, whether or not the progress of learning has improved after the warning. Whether or not the learning progress has improved is whether or not the average score of the comprehension test in a predetermined period after the warning is higher than the average score of the comprehension test in the predetermined period before the warning. Judged by criteria. Alternatively, this predetermined criterion may be determined based on whether or not the learning progress rate in the predetermined period after the warning is higher than the learning progress rate in the predetermined period before the warning. The learning progress rate is a parameter indicating the degree of learning progress per hour, specifically (number of chapters that passed the comprehension test) / day, or (video playback time of chapters that passed the comprehension test). / Day.

The timing determination means 14 determines the warning timing using the trained model thus obtained. For example, the timing determining means 14 generates a plurality of warning timing candidates according to a predetermined rule. The timing determination means 14 sequentially inputs these plurality of candidates into the trained model together with the past attendance records and profiles of the target students. For this input, the trained model outputs the expected result, that is, the possibility of obtaining the effect of the warning. The timing determining means 14 selects a plurality of candidates whose possibility of obtaining the effect of the warning exceeds the reference value, and determines the timing at which the warning is given. The reference value is, for example, predetermined (80%, etc.). Alternatively, this reference value may be determined by the manager or the student.

By determining the warning timing using machine learning in this way, it is possible to give a warning at an optimized timing based on the past achievements of the students.

The timing determination means 14 may set a plurality of model schedules determined by different methods for one user. For example, the timing determination means 14 may set two model schedules, a model schedule created by the administrator and a model schedule determined by machine learning, for one user.

3-3. Modification 3 (target of warning)
The method for extracting the students who are the target of the warning is not limited to the one exemplified in the embodiment. The students who are subject to the warning are, for example, (a) among the students whose difference between the student's attendance record and the model schedule meets the predetermined condition (first condition), (b) the student's attendance. Students whose achievements themselves meet certain conditions (second condition) (not comparison with the model schedule).

The first condition is, for example, at least one of the following.
(1a) The student's attendance record is behind the model schedule.
(1b) The delay from the model schedule of the student's attendance record is larger than the threshold value.
This threshold may be determined, for example, depending on the method for determining the model schedule. For example, if the model schedule was created by an administrator, the threshold was relatively small (ie, warnings were likely to occur), and the model schedule was determined using the attendance records of multiple users belonging to the user group. If it is, the threshold value is set relatively large (that is, it is difficult to issue a warning).

The second condition is, for example, at least one of the following.
(2a) The time change of the student's own attendance record is smaller than the threshold value.
(2b) The absolute value of the student's own attendance record is smaller than the threshold value.
(2c) The time change of the difference between the student's own attendance record and the representative value of the attendance record of other plurality of users belonging to the same user group is larger than the threshold value.
The threshold values of (2a) to (2c) may be determined, for example, according to a method for determining a model schedule. In one example, the learning support system 1 has a mathematical formula that calculates a threshold value according to the number of days elapsed from the start of the course. The learning support system 1 calculates the threshold value according to this mathematical formula. A plurality of coefficient sets in these formulas or formulas are prepared, and the learning support system 1 selects one formula or coefficient set according to the method for determining the model schedule. For example, if the model schedule was created by an administrator, the threshold was relatively small (that is, warnings were likely to occur), and the model schedule was determined using the attendance records of multiple users belonging to the user group. If it is, a formula or coefficient set is selected so that the threshold value is relatively large (that is, the warning is hard to be issued).

Alternatively, the threshold values of (2a) to (2c) may be determined using the statistical representative values of the attendance records of a plurality of users belonging to the user group. For example, the threshold value in (2a) is the value of the lower 20% in the distribution of the latest time change (for example, the increase in the attendance record in the last week) of the attendance record of a plurality of users belonging to the user group.

The threshold value in the second condition is not limited to the one using the statistical representative value of the attendance record in the user group to which the student belongs. A statistical representative value of a user group different from the user group to which the student belongs, for example, a user group consisting of students who have taken the same course in the past (hereinafter referred to as "past group") may be used. In this example, the warning means 15 extracts the students who are the target of the warning in comparison with the past group. For example, the warning means 15 specifies, at the time of determining whether or not to warn a certain student, the attendance record representing the time point corresponding to the time point in the past group. "Attendance record representing a certain point in time" means, for example, a group of students who have obtained the desired final result (specifically, have passed the target qualification test) and have a predetermined relative progress (as an example, the most). (Progress was slow) Refers to the attendance record of the students. Specifically, consider a case where it is determined whether or not to give a warning to a certain student at the time of 10 days until the end date of the course. Among the students who took this course in the last 3 years and passed the qualification test, even those who made the slowest progress in 10 days until the end of the course, 25 out of 50 chapters in 4 subjects If the chapter has been completed, the "Attendance record representing the remaining 10 days of the past group" is "Chapter 25/50". Next, the warning means 15 sets the attendance record of the student (for example, 20 chapters out of 50 chapters in 4 subjects have been taken and is behind the model schedule) at the time of the remaining 10 days in the past group. Contrast with the representative attendance record. If the difference between the two meets the criteria (for example, the attendance record of the student is behind the attendance record representing the past), the warning means 15 determines the student as the target of the warning.

The warning means 15 may extract the students who are the target of the warning regardless of the above-mentioned first condition and second condition. The warning means 15 may give a warning to a student who is not behind the model schedule but who meets other conditions, in addition to a person whose attendance record is behind the model schedule. A student who satisfies other conditions is, for example, a person whose attendance record is behind the attendance record representing a predetermined user group. The predetermined user group is, for example, a group of students who are currently taking the same course, or a group of students who have taken the same course in the past. The attendance record representing the user group refers to the attendance record at a predetermined position (for example, average value, top 10%, bottom 10%, etc.) in the distribution of the attendance record. A user group consisting of past students is a group of past students who meet specific conditions, such as those who have passed the target test or scores of various tests (for example, ability test) are above the reference point. It may be a user group consisting only of the students who were.

For example, the target of the warning may be determined using machine learning technology. The learning support system 1 is a target for which the data of the attendance schedule recorded as the attendance record of a plurality of attendees who have taken the course in the past, the profile of the attendee, and the warning are given to this machine learning model. And the timing is given to the input layer, and data indicating the presence or absence of the effect after the warning is given to the output layer as teacher data to perform machine learning.

The warning means 15 determines the target of the warning using the trained model thus obtained. For example, the warning means 15 generates a plurality of candidates to be warned according to a predetermined rule. The warning means 15 sequentially inputs these plurality of candidates into the trained model together with the past attendance records and profiles of the target students. For this input, the trained model outputs the expected result, that is, the possibility of obtaining the effect of the warning. The effectiveness of the warning is determined, for example, by whether or not the percentage of students expected to achieve the goal (eg, passing the qualification test) exceeds the reference value. The warning means 15 selects a plurality of candidates whose proportion of students who are expected to improve the progress of learning exceeds the reference value, and determines the target of the warning. The reference value is, for example, predetermined (80%, etc.). Alternatively, this reference value may be determined by the manager or the student.

By determining the target of the warning using machine learning in this way, it is possible to give a warning to the target optimized based on the past achievements of the students.

Further, the learning support system 1 may set a plurality of model schedules for one student. For example, the learning support system 1 sets two model schedules, a model schedule created by using a rule base and a model schedule created by an administrator, for one student. The warning means 15 extracts the target person of the warning for these two model schedules. In this case, the warning means 15 may set a partner to output a warning for each model schedule. For example, the warning means 15 outputs the warning about the student extracted in comparison with the model schedule created by using the rule base to the student himself / herself, and extracts it in comparison with the model schedule created by the administrator. Warnings about students are output to the administrator.

3-4. Modification 4 (contents of warning)
The content of the warning given by the warning means 15 is not limited to that exemplified in the embodiment. The warning means 15 may, for example, send a warning including a comparison with another student to the student. The comparison with other students includes, for example, the average value of the attendance record of a predetermined user group, the average score of the comprehension test, and the like. Alternatively, the warning means 15 may send a prediction of the final result based on the attendance record at that time to the student. Predictions of the final result can be obtained, for example, from a trained model. This warning includes, for example, the message "If you continue learning at this pace, the pass rate is 10%." Further, the warning means 15 may include information suggesting the content that the student should focus on learning. The content that should be focused on is, for example, the subject that the student is not good at. Alternatively, the content that should be focused on is the subject whose progress is delayed in comparison with other students.

Further, when the warning means 15 repeatedly gives a warning, the level may be changed each time the warning is given. In this case, the warnings are pre-leveled. For example, low-level warnings include messages of encouragement, medium-level warnings include contrasts with other students, and high-level warnings include predictions of final outcomes based on current attendance performance.

For example, the content of the warning may be determined using machine learning technology. The learning support system 1 gives the machine learning model the data of the course schedule recorded as the course record of a plurality of students who have taken the course in the past, the profile of the student, and the timing at which the warning is given. And the contents are given to the input layer, and the data indicating the presence or absence of the effect after the warning is given to the output layer as teacher data to perform machine learning.

The warning means 15 determines the warning content using the trained model thus obtained. For example, the warning means 15 generates a plurality of candidates for warning content according to a predetermined rule. The warning means 15 sequentially inputs these plurality of candidates into the trained model together with the past attendance records and profiles of the target students. For this input, the trained model outputs the possibility that the warning effect can be obtained as an expected result. The warning means 15 selects a plurality of candidates whose possibility of obtaining the effect of the warning exceeds the reference value, and determines the content of the warning. The reference value is, for example, predetermined (80%, etc.). Alternatively, this reference value may be determined by the manager or the student.

By determining the content of the warning using machine learning in this way, it is possible to give a warning with the content optimized based on the past achievements of the students.

Further, in addition to the warning automatically transmitted by the warning means 15, the administrator may manually send a warning or encouragement message.

Further, the warning means 15 is a user to a specific user (for example, an administrator) regardless of whether or not there is a student whose attendance record is behind the model schedule or whether or not there is a student who is the target of the warning. You may send the analysis result of the attendance record at that time for the entire group. This analysis result shows, for example, the distribution of the attendance record in the user group (corresponding to Fig. 8), the statistical representative value (maximum value, minimum value, average value, etc.), and the attendance record of the individual student who is the target of the warning. include.

3-5. Modification 5 (learning content)
The content of the learning content is not limited to that exemplified in the embodiment. The providing means 17 may adaptively change the content of the learning content according to the attendance record. For example, the providing means 17 increases the number of questions in the comprehension test or lowers the level of questions for subjects that the student is not good at based on the attendance record and subjects with relatively poor grades in the student group. .. Alternatively, the providing means 17 analyzes cross-subject (or cross-disciplinary) abilities (for example, memory ability, calculation ability, reading comprehension ability, etc.) from the results of the comprehension test, and has a problem for strengthening, which is a weak point. May be provided.

The providing means 17 omits the provision of some chapters according to the remaining period of the course (that is, the period until the scheduled end date), repeatedly provides at least some chapters, or changes the order of provision. You may.

3-6. Modification 6 (update of model schedule and warning timing)
The learning support system 1 may update the model schedule and / or the warning timing once determined. The model schedule and / or the warning timing is updated in the wake of a predetermined event. In one example, the model schedule may be updated according to the warning timing. For example, when the warning judgment is made weekly and it is decided to give a warning to a certain student, the schedule determination means 13 uses the model schedule for the user as the attendance record at that time. Update accordingly. For example, if there are 10 days left until the scheduled end date at the time of issuing a warning and 30 chapters have not been learned, the schedule determination means 13 determines a model schedule for learning 30 chapters in the remaining 10 days. The warning means 15 sends an updated model schedule to the students together with the warning. From the next time onward, the warning means 15 determines whether or not to warn the student in comparison with the updated model schedule. Examples of the event that triggers the update of the model schedule and / or the warning timing include an update instruction from the student or the administrator, an event on the calendar (for example, the event schedule has changed), and the like.

3-7. Other Modifications Although examples of using machine learning have been described in

Modifications

1, 2, 3, and 4, the teacher data exemplified in these examples may be used in combination. For example, the timing and content of warnings may be optimized using machine learning.

The structure of the course is not limited to that illustrated in the embodiment. For example, the target course may be single.

The learning support system 1 may have a function of editing (or changing) the model schedule determined by the schedule determination means 13. Similarly, the learning support system 1 may have a function of editing the warning timing determined by the timing determination means 14. These edits are made, for example, by the student's administrator.

The specific device configuration of the learning support system 1 is not limited to that exemplified in FIG. The learning support system 1 may have any device configuration as long as it can implement the required functions. For example, a part of the functions described as the functions of the management server 10 in the embodiment may be implemented in a device different from the management server 10, for example, the client 20. Further, the management server 10 may be a physical server or a virtual server (including a so-called cloud).

The program executed by the CPU 110 may be provided in a state of being recorded on a computer-readable recording medium, for example, a DVD-ROM, or may be downloaded from a server on the Internet.

Claims

Specific means of identifying online courses and
A schedule determination means for determining a model schedule for taking the online course without using the attendance results of users who have taken the online course in the past.
A timing determining means for determining the timing of warnings for the model schedule, and
A learning support system having a warning means for giving a warning at the timing to a user extracted from users whose attendance record is behind the model schedule.
The learning support system according to claim 1, wherein the warning means warns a user whose attendance record change rate is lower than the standard.
The learning support system according to claim 1 or 2, wherein the warning means gives a warning to a user whose course progress is relatively slow in a user group including a plurality of users.
The learning support system according to claim 3, wherein the user group is a group consisting of users who have taken the online course in the past or a group consisting of users who are actually learning according to the same model schedule.
The online course is divided into multiple chapters.
The online course has a comprehension test after at least some chapters and
The learning support system according to any one of claims 1 to 4, wherein the attendance record is specified according to the answer to the comprehension test.
The learning support system according to any one of claims 1 to 5, wherein the attendance record is specified according to the time taken for the online course, or the progress rate or the enrollment rate of the online course.
The online course includes multiple subjects
The learning support system according to any one of claims 1 to 6, wherein the attendance record is specified according to the weight given to each of the plurality of subjects.
The schedule determination means is any one of claims 1 to 7 for determining the model schedule using a machine learning model trained by giving the attendance record and the final result of the user who has taken the online course in the past as teacher data. The learning support system described in item 1.
The timing determination means determines the model schedule using a machine learning model trained by giving the attendance record of a user who has taken the online course in the past, the timing of warning, and the effect as teacher data. The learning support system according to any one of Items 1 to 8.
The warning means is a user who satisfies the conditions determined by using a machine learning model trained by giving the attendance record of the user who has taken the online course in the past, the target to which the warning has been given, and the effect as teacher data. The learning support system according to any one of claims 1 to 9, which is extracted as a warning target.
The warning means determines the content of the warning using a machine learning model trained by giving the attendance record of the user who has taken the online course in the past, the content of the warning, and the effect as teacher data. The learning support system according to any one of 10.
On the computer
Steps to identify online courses and
A step to determine the model schedule for taking the online course without using the attendance results of users who have taken the online course in the past.
Steps to determine the timing of warnings for the model schedule,
A program for executing a step of giving a warning at the timing to a user extracted from users whose attendance record is behind the model schedule.
The steps by which the computer identifies the online course,
A step in which the computer determines a model schedule for taking the online course without using the results of the users who have taken the online course in the past.
A step in which the computer determines the timing of warnings for the model schedule.
A learning support method in which the computer has a step of giving a warning at the timing to a user extracted from users whose attendance record is behind the model schedule.