JP7694068B2 - Comment output device, comment output program, learning support system, and comment output method - Google Patents

Description

The present invention relates to a comment output device, a comment output program, a learning support system, and a comment output method for outputting comments to learners.

Efforts have been made to support learners in their learning. For example, a learning support device has been disclosed that provides appropriate learning support to individual learners when they study content using analog learning materials such as paper drills or digital learning materials such as tablets (for example, Patent Document 1).

JP 2018-59973 A

The technique described in Patent Document 1, for example, is intended to enable a learner to study efficiently by using edited questions for mastery learning materials and questions for evaluation materials depending on the learner's level of comprehension of the learning content.
A common learning support method is to suggest learning materials based on the learner's level of understanding, as described in Patent Document 1. However, no matter how many learning materials are suggested to a learner, if the learner does not have the motivation to learn, it will not lead to improvement in academic ability. Therefore, it is also important to improve the learner's motivation to learn and maintain their motivation, such as by encouraging or giving advice to improve the learner's motivation to learn.

The present invention aims to provide a comment output device, a comment output program, a learning support system, and a comment output method for improving learners' motivation to study.

The present invention solves the above problems by the following solving means.
A first invention is a comment output device comprising: learning achievement information acquisition means for acquiring learning achievement information of a learner; comment output means for selecting and outputting a comment regarding learning for the learner from a plurality of comments stored in a comment storage unit; and learning means for performing reinforcement learning regarding the selection of the comment based on the learning achievement information before the comment is output by the comment output means and the learning achievement information after the comment is output.
A second invention is a comment output device according to the first invention, further comprising a comment selection means for selecting a comment from the plurality of comments stored in the comment memory unit based on a value function updated by the learning means through reinforcement learning, and the comment output means outputs the comment selected by the comment selection means.
A third invention is a comment output device according to the second invention, wherein the comment storage unit stores the comments in a plurality of categories, and the learning means performs reinforcement learning related to the selection of the comments for each category.
A fourth invention is a comment output device in which, in any of the first to third inventions, the learning achievement information includes information relating to learning achievement, and the learning means performs reinforcement learning to give a reward to the comment if the learning achievement has increased as a result of comparing the learning achievement after the comment is output with the learning achievement before the comment is output.
A fifth invention is a comment output device in which, in any of the first to fourth inventions, the learning achievement information includes information relating to learning time, and the learning means performs reinforcement learning to give a reward to the comment when the learning time has increased as a result of comparing the learning time after the comment is output with the learning time before the comment is output.
A sixth invention is a comment output device according to the fourth or fifth invention, wherein the learning means determines the magnitude of the reward depending on the degree of change between the learning achievement information after the comment is output and the learning achievement information before the comment is output.
A seventh invention is a comment output device in any of the fourth to sixth inventions, wherein the learning means determines the magnitude of the reward depending on the learning level of the learner before outputting the comment.
The eighth invention is a comment output device which is any of the first to seventh inventions, and which is provided with a learner information acquisition means for acquiring learner information which is at least one of the learner's attribute information and personality tendencies, and the learning means performs reinforcement learning related to the selection of the comment for each of the learner information acquired by the learner information acquisition means.
A ninth invention is a comment output device which is any of the comment output devices from the first to eighth inventions, and which comprises a question sentence memory unit which stores question sentences related to learning, and a question sentence output means which extracts and outputs the question sentence from the question sentence memory unit based on the learning achievement information acquired by the learning achievement information acquisition means.
A tenth aspect of the present invention is a comment output program for causing a computer to function as any one of the comment output devices according to the first to ninth aspects of the present invention.
An eleventh invention is a learning support system in which a comment output device and a learner terminal are communicatively connected, the learner terminal comprising an achievement information sending means for transmitting learning achievement information of the learner, the comment output device comprising: learning achievement information acquisition means for acquiring the learning achievement information from the learner terminal; comment output means for selecting a comment regarding learning for the learner from a plurality of comments stored in a comment storage unit and outputting the comment to the learner terminal; learning means for performing reinforcement learning related to the selection of the comment based on the learning achievement information acquired by the learning achievement information acquisition means before the comment is output by the comment output means and the learning achievement information acquired by the learning achievement information acquisition means after the comment is output; and comment selection means for selecting the comment from the plurality of comments stored in the comment storage unit based on a value function updated as a result of the learning means performing reinforcement learning, and the comment output means outputs the comment selected by the comment selection means to the learner terminal.
A twelfth invention is a comment output method including: a learning achievement information acquisition step in which a computer acquires learning achievement information of a learner; a comment output step in which a comment regarding learning for the learner is selected from a plurality of comments stored in a comment storage unit and output; and a learning step in which reinforcement learning related to the selection of the comment is performed based on the learning achievement information before the comment is output by the comment output step and the learning achievement information after the comment is output.

The present invention provides a comment output device, a comment output program, a learning support system, and a comment output method for improving learners' motivation to study.

1 is a functional block diagram of a learning support system according to an embodiment of the present invention. FIG. 4 is a diagram illustrating an example of a comment storage unit of a comment output server according to the embodiment. 13 is a flowchart showing a comment output process in the comment output server according to the embodiment. 13 is a flowchart showing a learning process in the comment output server according to the embodiment. 11A and 11B are diagrams showing specific examples of learning achievement levels and comment selection according to the embodiment; 11A and 11B are diagrams showing specific examples of learning achievement levels and comment selection according to the embodiment; 11A and 11B are diagrams showing specific examples of learning achievement levels and comment selection according to the embodiment;

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Note that this is merely an example, and the technical scope of the present invention is not limited to this example.
(Embodiment)
<Overall configuration of learning support system 100>
FIG. 1 is a functional block diagram of a learning support system 100 according to this embodiment.
FIG. 2 is a diagram showing an example of the comment storage unit 22 of the comment output server 1 according to this embodiment.

1 is a system in which a comment output server 1 (comment output device) acquires learning achievement information from a terminal 5 (learner terminal) and outputs a comment related to learning to the terminal 5. The comment related to learning is selected based on a value function updated by the comment output server 1 through reinforcement learning regarding the relationship between the output comment and the learning achievement information before and after the output of the comment.
The learning support system 100 includes a comment output server 1 and a plurality of terminals 5. The comment output server 1 and the terminals 5 are connected via a communication network N.

<Comment output server 1>
The comment output server 1 is, for example, a server managed by a company that provides analog or digital learning materials, or a service provider that supports learners' learning.
The comment output server 1 receives, for example, learning achievement data (learning achievement information) from the learner's terminal 5. Then, the comment output server 1 outputs a comment based on the received learning achievement data to the learner's terminal 5. Since the value function due to learning has not been updated initially, the comment output by the comment output server 1 may be any comment, and may be, for example, a comment randomly selected from comments stored in a comment storage unit 22 (described later).

Next, the comment output server 1 receives the learning achievement data again from the terminal 5 of the learner who output the comment. The learning achievement data here is the learning achievement data after the comment is output, and is therefore considered to have been influenced by the comment to some extent.
Then, the comment output server 1 learns the changes obtained from the learning achievement data before and after the output of the comment on the comment.
Moreover, the comment output server 1 outputs a comment based on the received learning achievement data to the learner's terminal 5. Since the value function has been updated by the learning, the comment output server 1 selects and outputs a comment from the comment storage unit 22 based on the value function.

The comment output server 1 includes a control unit 10 , a storage unit 20 , and a communication interface unit 29 .
The control unit 10 is a CPU (Central Processing Unit) that controls the entire comment output server 1. The control unit 10 appropriately reads and executes an OS (Operating System) and various application programs stored in the storage unit 20, thereby cooperating with the above-mentioned hardware and executing various functions.
The control unit 10 includes a learning achievement information acquisition unit 11 (learning achievement information acquisition means), a comment selection unit 12 (comment selection means), a comment output unit 13 (comment output means), a learning unit 14 (learning means), and a question statement output unit 15 (question statement output means).

The learning achievement information acquiring unit 11 receives the learning achievement data transmitted by the learner's terminal 5. Here, the learning achievement data indicates, for example, the learning proficiency level, and may be a test score, a deviation value, an evaluation value, or the like. The learning achievement data may also be, for example, the study time. In the case of the study time, for example, the terminal 5 measures the time the learner has studied using the terminal 5, and the total time the terminal 5 has been used may be the study time.
The comment selection unit 12 selects a comment from a plurality of comments stored in the comment storage unit 22. Here, each comment is, for example, a comment that encourages the learner and is intended to increase the learner's motivation to study.
The comment output unit 13 transmits the comment selected by the comment selection unit 12 to the terminal 5 .

The learning unit 14 performs learning related to the selection of a comment based on the learning achievement data of the learner before the comment is output and the learning achievement data of the learner after the comment is output.
The learning unit 14 performs learning using, for example, reinforcement learning.
Here, the basic mechanism of reinforcement learning will be described. An agent (corresponding to the comment output server 1 in this embodiment) acquires the state of the environment (corresponding to the learner in this embodiment), selects an action, and the environment changes based on the action. As the environment changes, some kind of reward is given, and the agent learns to select better actions (decision making).

In reinforcement learning, by learning actions, the robot learns appropriate actions based on the interactions that the actions have with the environment, that is, it learns how to learn in a way that maximizes future rewards. In this embodiment, this means that the robot can acquire actions that will have an impact on the future, such as selecting a comment, which is behavioral information that will further improve the learning achievement data, which is a state.

Reinforcement learning is
(1) Action;
(2) Feedback of changes in the environment and rewards due to behavior,
(3) Compare the value of taking that action with the value of not taking that action, revise your strategy, and go back to (1).
Learning continues by repeating this cycle.

When the above cycle is applied to this embodiment, the learning unit 14 performs the following:
(1) Select a comment and output it to the learner.
(2) If the change in learning achievement data due to a comment indicates an improvement in learning achievement, a larger reward will be given to the comment.
(3) Update the value function to maximize the learning achievement.
This cycle is repeated as we learn.
Then, learning unit 14 stores the updated value function in value function storage unit 25 (described later).

The question output unit 15 extracts a question from the question storage unit 23 (described later) based on the learning achievement data received by the learning achievement information acquisition unit 11, and transmits it to the learner's terminal 5.

The storage unit 20 is a storage device such as a hard disk or a semiconductor memory device for storing programs, data, etc. required for the operation of the comment output server 1 .
Note that a computer refers to an information processing device equipped with a control unit, a storage device, etc., and the comment output server 1 is an information processing device equipped with a control unit 10, a storage unit 20, etc., and is included in the concept of a computer.
The memory unit 20 includes a program memory unit 21 , a comment memory unit 22 , a question statement memory unit 23 , and a value function memory unit 25 .

The program storage unit 21 is a storage area for storing programs. The program storage unit 21 stores a comment output program 21a. The comment output program 21a is a program for executing various functions of the control unit 10. Note that instead of the comment output program 21a, the control unit 10 may have a function for outputting comments and a function for learning as separate programs.

The comment storage unit 22 is a storage area for storing comments regarding learning to be output to the learner.
2, the comment storage unit 22 stores a number and a comment content in association with each other. The comment storage unit 22 classifies comments into, for example, a group of comments 22a for when the learning achievement level has improved and a group of comments 22b for when the learning achievement level has decreased.

The question storage unit 23 is a storage area for storing question sentences. The question storage unit may store question sentences of different levels of difficulty in association with the learning achievement levels.
The value function storage unit 25 is a storage area that stores the value function updated by the learning unit 14. The value function stored in the value function storage unit 25 is updated by the learning unit 14.
The communication interface unit 29 is an interface unit for communicating with the terminal 5 via the communication network N.

There is no limit to the number of pieces of hardware that make up the comment output server 1. The number of pieces of hardware that make up the comment output server 1 may be one or more, as necessary. Furthermore, the hardware of the comment output server 1 may be configured to include various servers such as a Web server, a DB (database) server, an application server, and a learning server, as necessary, and may be configured as one server or as separate servers.
Furthermore, the comment storage unit 22, the question sentence storage unit 23, and the value function storage unit 25 stored in the storage unit 20 of the comment output server 1 may be provided in a separate server. In that case, the control unit 10 of the comment output server 1 may obtain or update various data stored in the separate server by communicating with the separate server.

<Terminal 5>
The terminal 5 is a terminal used by a learner. The terminal 5 is, for example, a tablet terminal. The terminal 5 is not limited to a tablet terminal, and may be a PC (personal computer), a mobile terminal, or the like. The terminal 5 may also include a terminal used by a teacher.
Although not shown, the terminal 5 includes a control unit, a storage unit, an input unit, a display unit, a communication interface unit, etc. In the case of a tablet terminal, the input unit and the display unit of the terminal 5 are integrated into a touch panel display.
Further, a computer refers to an information processing device equipped with a control unit, a storage device, etc., and the terminal 5 is an information processing device equipped with a control unit, a storage device, etc., and is included in the concept of a computer.

The communication network N is a network between the comment output server 1 and the terminal 5, and is a communication network such as an Internet line. The communication network N may be either wired or wireless.

Next, the processing performed by the learning support system 100 will be described.
<Comment output processing>
FIG. 3 is a flowchart showing a comment output process in the comment output server 1 according to this embodiment.
For example, it will be described that on terminal 5, a learner uses a learning app (not shown), which is software for studying that is stored in terminal 5, to study on terminal 5 for each unit, and takes a confirmation test at the end of the unit.

When the learner takes the confirmation test using the terminal 5 and, for example, executes a marking process, the control unit (achievement information transmitting means) of the terminal 5 transmits learning achievement data.
Then, in step S (hereinafter, “step S” is simply referred to as “S”) 11 in FIG. 3, the control unit 10 (learning achievement information acquisition unit 11) of the comment output server 1 receives the learning achievement data from the terminal 5.

In S<b>12 , the control unit 10 (comment output unit 13 ) randomly selects one comment from the comments stored in the comment storage unit 22 , and outputs it to the terminal 5 .
The learner studies the next unit by looking at the comments displayed on the terminal 5. After the learner takes a confirmation test for the next unit using the terminal 5, the control unit (achievement information transmission means) of the terminal 5 executes the same process as above to transmit the learning achievement data.
Then, in S<b>13 , the control unit 10 (learning achievement information acquisition unit 11 ) receives the learning achievement data from the terminal 5 .

In S14, the control unit 10 (the learning unit 14) performs a learning process to update the value function. The learning process will be described later.
In S<b>15 , the control unit 10 (comment selection unit 12 ) selects a comment using the value function stored in the value function storage unit 25 .

In S16, the control unit 10 judges whether the learning achievement level of the received learning achievement data is equal to or lower than a predetermined value. Here, when the learning achievement level is equal to or lower than the predetermined value, it means that the understanding of the learning has not progressed. When the learning achievement level is equal to or lower than the predetermined value (S16: YES), the control unit 10 transfers the process to S17. On the other hand, when the learning achievement level is not equal to or lower than the predetermined value (S16: NO), the control unit 10 transfers the process to S19.

In S17, the control unit 10 (question sentence output unit 15) extracts a question sentence from the question sentence storage unit 23 based on the learning achievement data received in S13.
In S18, the control unit 10 (comment output unit 13, question sentence output unit 15) transmits the comment selected in S15 and the question sentence extracted in the process of S17 to the terminal 5. Thereafter, the control unit 10 shifts the process to S13, receives the learning achievement data from the terminal 5, and repeats the process of outputting the comment.
On the other hand, in S19, the control unit 10 (comment output unit 13) transmits the comment selected in S15 to the terminal 5. Thereafter, the control unit 10 shifts the process to S13 and repeats the process.

<Learning process>
Next, a learning process related to reinforcement learning will be described.
FIG. 4 is a flowchart showing the learning process in the comment output server 1 according to this embodiment.
In S21, the control unit 10 (learning unit 14) compares the learning achievement level before and after the comment is sent.
In S22, the control unit 10 (learning unit 14) determines whether the learning achievement level has improved. The control unit 10 determines the result of the learning achievement level after the comment is sent based on the learning achievement level before the comment is sent. If the learning achievement level has improved (S22: YES), the control unit 10 moves the process to S23. On the other hand, if the learning achievement level has not improved (S22: NO), the control unit 10 moves the process to S26. If the learning achievement level has not changed, this may include, for example, a case where the learning achievement level has not improved.

In S23, the control unit 10 (learning unit 14) determines whether the learning achievement level (learner's learning level) before the comment was sent was equal to or greater than a predetermined value. If the learning achievement level before the comment was sent was equal to or greater than the predetermined value (S23: YES), the control unit 10 transfers the process to S24. On the other hand, if the learning achievement level before the comment was sent was not equal to or greater than the predetermined value (S23: NO), the control unit 10 transfers the process to S25.

In S24, the control unit 10 (the learning unit 14) sets the reward to a first value. Here, the first value is a positive value. After that, the control unit 10 shifts the process to S27.
In S25, the control unit 10 (the learning unit 14) sets the reward to a second value. Here, the second value is a positive value and is smaller than the first value. After that, the control unit 10 moves the process to S27.
In S26, the control unit 10 (the learning unit 14) sets the reward to a third value. Here, the third value is a negative value.
In S27, the control unit 10 (learning unit 14) updates the value function based on the reward. After that, the control unit 10 ends this process and moves the process to S15 in FIG.

<Specific examples>
Next, specific examples of the learning achievement level and the comments will be described.
5 to 7 are diagrams showing specific examples of the learning achievement level and comment selection according to the present embodiment.
5A shows the learning achievement of a certain learner (student A) as a graph 60. The graph 60 shows a past period 61 including the current time, a current achievement level 62, a past achievement level 63 including the current time, and a future progress 64 of the achievement level that is undetermined at the time of the current achievement level 62.
FIG. 5B shows a comment 221 that is output when the current achievement level is 62.
If the learning achievement level after outputting the comment 221 increases as shown in future trend 64, the learning unit 14 sets the reward value for number "C001" (see Figure 2) in the comment memory unit 22 to a positive value and updates the value function.

6A shows the learning achievement of a certain learner (student B) as a graph 70. The graph 70 shows a past period 71 including the current time, a current achievement level 72, a past achievement level 73 including the current time, and a future progress 74 of the achievement level that is undetermined at the time of the current achievement level 72.
FIG. 6B shows a comment 222 that is output when the current achievement level is 72.
If the learning achievement level after outputting the comment 222 decreases as shown in future trend 74, the learning unit 14 sets the reward value for number "C102" (see Figure 2) in the comment memory unit 22 to a negative value and updates the value function.

7A shows the learning achievement of a certain learner (student C) as a graph 80. The graph 80 shows a past period 81 including the current time, a current achievement level 82, a past achievement level 83 including the current time, and a future progress 84 of the achievement level that is undetermined at the time of the current achievement level 82.
FIG. 7B shows a comment 223 to be output when the current achievement level is 82.
If the learning achievement level after outputting the comment 223 decreases as shown in future trend 84, the learning unit 14 sets the reward value for number "C101" (see Figure 2) in the comment memory unit 22 to a negative value and updates the value function.

As described above, according to this embodiment, the comment output server 1 has the following advantages.
(1) The learner's learning achievement data is obtained, and comments related to learning are output to the learner's terminal 5. Reinforcement learning is performed for selecting comments based on the learning achievement level before the comment is output and the learning achievement level after the comment is output, so that changes in the learner's achievement level in response to the comments can be learned.
In addition, a comment is selected from the multiple comments stored in the comment storage unit 22 based on the value function updated by the reinforcement learning.
As a result, it is possible to output to the learner a comment that corresponds to the learning achievement level and enhances the learner's motivation for learning.

(2) By comparing the learning achievement after the comment is output with the learning achievement before the comment is output, reinforcement learning is performed to give a higher reward when the learning achievement increases. Therefore, it is possible to set a higher reward for comments output before the learning achievement increases.

(3) When the learning achievement level is below a predetermined value, a question is extracted from the question storage unit 23 and output to the terminal 5. Therefore, for learners with a low learning achievement level, the question can be output to encourage them to improve their learning achievement level.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above-mentioned embodiments. Furthermore, the effects described in the embodiments are merely a list of the most favorable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments. Note that the above-mentioned embodiments and the modified forms described below can be used in appropriate combinations, but detailed explanations will be omitted.

(Modifications)
(1) In the present embodiment, the learning achievement data is acquired by receiving it from the learner's terminal, but this is not limiting. For example, a teacher may input the learning achievement data of the learner from the teacher's terminal, and the control unit of the comment output server may acquire the learning achievement data.
Similarly, although the above description is given with an example in which the comments are sent to the terminal of the learner, the present invention is not limited to this. For example, the comments may be output to the terminal of the teacher, and the teacher may provide guidance to the learner based on the output comments.

(2) In the present embodiment, the learning related to the comment selection using the learning achievement level has been described, but the present invention is not limited to this. For example, the learning time may be used instead of or in addition to the learning achievement level. Then, the control unit of the comment output server may provide a reward when the learning time increases.
Furthermore, the size of the reward may be determined according to the degree of change before and after the comment is output and the level of the learner before the comment is output. For example, even if the degree of change before and after the comment is output is small, if the level of the learner before the comment is output is high (the learner was originally a good learner), the reward may be large. On the other hand, if the degree of change before and after the comment is small and the level of the learner before the comment is low, the reward may be small.

(3) In the present embodiment, one comment storage unit is prepared and divided into categories that reflect the learning results. However, this is not limited to this. For example, the comment output server may divide the comment storage unit into categories such as by subject or by learner group (elementary school students, junior high school students, etc.).

(4) In this embodiment, the learners are not particularly limited, but the present invention is not limited to this. For example, learning may be performed separately for each learner information, such as the learner's attribute information and personality tendencies. The attribute information may be, for example, age or gender, and the personality tendencies may be learning tendencies, such as whether a learner improves when praised or when scolded. In this case, the control unit (learner information acquisition means) of the comment output server acquires the learner information, for example, from the learner via a terminal.

(5) In the present embodiment, a question text is output when the learning achievement level is equal to or lower than a predetermined value, but the present invention is not limited to this. A question text may be output regardless of the learning achievement level. Furthermore, a question with a level of difficulty according to the learning achievement level may be output.

(6) In this embodiment, the results of a confirmation test for each unit are used as learning achievement data, but this is not limiting. For example, it may be correct/incorrect data on questions for qualification exam preparation or study questions appropriate for a child's grade level, and anything related to learning may be used.

(7) In this embodiment, a learning app is used, but this is not limiting. For example, a web service may be used.

(8) In the present embodiment, the learning process and the comment output process are performed by the comment output server, but this is not limiting. The learning process and the comment output process may be performed by different devices.

REFERENCE SIGNS LIST 1 Comment output server 5 Terminal 10 Control unit 11 Learning achievement information acquisition unit 12 Comment selection unit 13 Comment output unit 14 Learning unit 15 Question statement output unit 20 Memory unit 21a Comment output program 22 Comment memory unit 23 Question statement memory unit 25 Value function memory unit 100 Learning support system N Communication network

Claims (12)

A learning achievement information acquiring means for acquiring learning achievement information including information related to a learner's learning achievement level ;
a question output means for extracting and outputting a question from a question storage unit that stores questions related to learning based on the learning achievement information when the learning achievement level of the learning achievement information acquired by the learning achievement information acquisition means is equal to or lower than a predetermined value;
a comment output means for selecting and outputting a comment regarding learning to the learner from among a plurality of comments stored in a comment storage unit;
a learning means for performing reinforcement learning related to selection of the comment based on the learning achievement information before the comment output means outputs the comment and the learning achievement information after the comment is output;
A comment output device comprising: 2. The comment output device according to claim 1,
a comment selection means for selecting the comment from the plurality of comments stored in the comment storage unit based on a value function updated by the learning means through reinforcement learning,
The comment output means outputs the comment selected by the comment selection means. 3. The comment output device according to claim 2,
The comment storage unit stores the comments by dividing them into a plurality of categories,
The learning means performs reinforcement learning regarding the selection of the comment for each of the categories. In the comment output device according to any one of claims 1 to 3 ,
The learning means of the comment output device performs reinforcement learning to give a reward to the comment if the learning achievement level has increased as a result of comparing the learning achievement level after the comment is output with the learning achievement level before the comment is output. In the comment output device according to any one of claims 1 to 4,
The learning achievement information includes information regarding learning time,
The learning means performs reinforcement learning to give a reward to the comment when the learning time has increased as a result of comparing the learning time after the comment is output with the learning time before the comment is output. 6. The comment output device according to claim 4,
The learning means determines the size of the reward according to a degree of change between the learning achievement information after the comment is output and the learning achievement information before the comment is output. In the comment output device according to any one of claims 4 to 6,
The learning means determines the size of the reward according to the learning level of the learner before the comment is output. In the comment output device according to any one of claims 1 to 7,
a learner information acquiring means for acquiring learner information, which is at least one of attribute information and personality tendency of the learner,
The learning means performs reinforcement learning regarding the selection of the comment for each of the learner information acquired by the learner information acquisition means. In the comment output device according to any one of claims 1 to 8,
The question output means extracts from the question storage unit and outputs the question having a difficulty level corresponding to the learning achievement level of the learning achievement information acquired by the learning achievement information acquisition means. A comment output program for causing a computer to function as a comment output device according to any one of claims 1 to 9. A learning support system in which a comment output device and a learner terminal are communicatively connected,
The learner terminal includes an achievement information transmitting means for transmitting learning achievement information including information regarding the learner's learning achievement level ,
The comment output device includes:
a learning achievement information acquiring means for acquiring the learning achievement information from the learner terminal;
a question output means for extracting a question from a question storage unit that stores questions related to learning based on the learning achievement information and outputting the question to the learner terminal when the learning achievement level of the learning achievement information acquired by the learning achievement information acquisition means is equal to or lower than a predetermined value;
a comment output means for selecting a comment regarding learning for the learner from a plurality of comments stored in a comment storage unit and outputting the selected comment to the learner terminal;
a learning means for performing reinforcement learning related to selection of the comment based on the learning achievement information acquired by the learning achievement information acquisition means before the comment is output by the comment output means and the learning achievement information acquired by the learning achievement information acquisition means after the comment is output;
a comment selection means for selecting a comment from the plurality of comments stored in the comment storage unit based on a value function updated by the learning means through reinforcement learning;
Equipped with
The comment output means outputs the comment selected by the comment selection means to the learner terminal. The computer
A learning achievement information acquisition step of acquiring learning achievement information including information related to the learning achievement level of the learner;
a question output step of extracting and outputting a question from a question storage unit that stores question sentences related to learning based on the learning achievement information when the learning achievement level of the learning achievement information acquired by the learning achievement information acquisition step is equal to or lower than a predetermined value;
a comment output step of selecting and outputting a comment regarding learning to the learner from among a plurality of comments stored in a comment storage unit;
a learning step of performing reinforcement learning related to selection of the comment based on the learning achievement information before the comment is output by the comment output step and the learning achievement information after the comment is output;
Comment output method, including

Images