WO2021177416A1 - Learning assistance device, learning assistance method, and program - Google Patents

Abstract

The present invention allows a user to receive education of good quality by means of a case method in an e-learning environment. A learning assistance device according to the present invention is provided with: a question setting unit that causes a question in a natural language to be set in a learner terminal device through communication with the learner terminal device via a network, and causes a feedback according to a learner's answer to the question to be presented to the learner in the learner terminal device; an answer analysis unit that performs a prescribed analysis with respect to the answer in the natural language to the question, the answer having been obtained from the learner terminal device through the communication via the network; and a determination unit that determines, on the basis of the analysis result of the answer analysis unit, the feedback to be presented by the question setting unit and the next question to be set by the question setting unit.

Description

Learning support device, learning support method and program

The present invention relates to a learning support device, a learning support method and a program.
The present application claims priority based on Japanese Patent Application No. 2020-037108 filed in Japan on March 4, 2020, the contents of which are incorporated herein by reference.

There is known a learning support device that collects answers to test items by multiple learners via a network and sends review questions according to the item characteristics of the test items analyzed based on the collected answers. (See, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2018-205354

An educational method called the case method is known. As an example, in the case of business administration, the case method is based on a case that actually occurred in a certain company, and if you are a manager, you will learn how to deal with that case. You can mention things like discussing with each other. In addition to business administration, case-method education is also regarded as important in fields such as law and medicine.
Further, in such a case method, the role of the instructor is not to give a lecture or teach one's own theory to the students, but to organize the discussions of the students and guide them in a preferable direction. In this respect, it is said that the educational effect of the case method largely depends on the skill of the instructor. For this reason, the case method is recognized as one in which the student actually learns in an environment such as a classroom where the instructor is present.
On the other hand, it is required that users can receive high-quality education by the case method under the e-learning environment. For this purpose, due to the nature of the case method, it is necessary to appropriately guide the learner's thoughts, etc. according to the individuality of the learner's personality, ability, etc., as is done by the actual instructor. Desired.
However, since the technique described in Cited Document 1 is premised on providing a common test item and a plurality of questions to a plurality of learners, it is applied to a case method and appropriately applied to each learner. It is difficult to guide.

In view of such a situation, the present invention aims to enable the user to receive high-quality education by the case method in an e-learning environment.

One aspect of the present invention for solving the above-mentioned problems is to allow a problem in natural language to be asked by the learner terminal device through communication with the learner terminal device via a network, and to solve the problem. For the question obtained from the learner terminal device via communication via a network and a questioning section that allows the learner to present feedback to the learner according to the learner's answer to the question. An answer analysis unit that analyzes answers in natural language, and a decision unit that determines the feedback presented by the question section and the next question to be asked by the question section based on the analysis results of the answer analysis unit. It is a learning support device equipped with.

Further, one aspect of the present invention is to allow a problem in natural language to be asked on the learner terminal device through communication with the learner terminal device via a network, and to provide a learner's answer to the problem. Analysis of the question step so that the corresponding feedback to the learner is presented on the learner terminal device, and the answer in natural language to the problem obtained from the learner terminal device via communication via the network. A learn support method including an answer analysis step for determining an answer analysis step, a feedback presented by the question step based on the analysis result of the answer analysis step, and a decision step for determining the next question to be asked by the question step. Is.

Further, one aspect of the present invention is to make a computer as a learning support device ask a problem in natural language on the learner terminal device via communication with the learner terminal device via a network. , The question section for presenting feedback to the learner according to the learner's answer to the problem on the learner terminal device, and the question unit obtained from the learner terminal device via communication via a network. Based on the answer analysis unit that analyzes the answer to the question in natural language and the analysis result by the answer analysis unit, the feedback presented by the question section and the next question to be asked by the question section are determined. It is a program to function as a decision unit.

As described above, according to the present invention, it is possible to obtain the effect that the user can receive sufficient education by the case method under the e-learning environment.

It is a figure which shows the overall configuration example of the learning support system of this embodiment. It is a figure which shows the configuration example of the learning support apparatus of this embodiment. It is a figure which shows an example of the problem information of this embodiment. It is a figure which shows the structural example of the problem group data of this embodiment. It is a flowchart which shows the example of the processing procedure executed by the user terminal apparatus and learning support apparatus of this embodiment.

FIG. 1 shows an overall configuration example of the learning support system of the present embodiment. The learning support system of the present embodiment supports the user U to perform learning by the case method (case method learning) using his / her own user terminal device 100. That is, the user U in the present embodiment is a person (learner, learner) who performs case method learning using the user terminal device 100.
The learning support system of the present embodiment includes a plurality of user terminal devices 100 and a learning support device 200.
The user terminal device 100 (an example of the learner terminal device) is a terminal device used by each user U to perform learning by the case method. The user terminal device 100 may be, for example, a personal computer, a tablet terminal, a smartphone, or the like.

The learning support device 200 is a device that provides the user U with learning support corresponding to case method learning by communicating with each of the user terminal devices 100 via the network NT. That is, the learning support device 200 provides the case method course to the user U by communicating with the user terminal device 100.

FIG. 2 shows a configuration example of the learning support device 200. The learning support device 200 shown in the figure includes a communication unit 201, a control unit 202, and a storage unit 203.
The communication unit 201 executes communication via the network NT.

The control unit 202 executes various controls in the learning support device 200. The function as the control unit 202 is realized by executing a program by a CPU (Central Processing Unit) included in the learning support device 200. The control unit 202 includes a question unit 221, an answer analysis unit 222, and a determination unit 223.
The questioning unit 221 allows the user terminal device 100 to ask a question through communication with the user terminal device 100 via the network NT, and provides feedback to the learner according to the learner's answer to the question. Is presented on the user terminal device 100.
The answer analysis unit 222 analyzes predetermined items for the answers obtained from the user terminal device 100 via communication via the network.
The decision unit 223 determines the feedback presented by the question unit 221 and the next question to be set by the question unit 221 based on the analysis result by the answer analysis unit 222.

The storage unit 203 stores various information related to the learning support device 200. The storage unit 203 includes a user information storage unit 231 and a problem information storage unit 232.
The user information storage unit 231 stores information for each user as a student who takes the case method course provided by the learning support device 200.
The problem information storage unit 232 stores problem information including problems prepared for each case as a subject.

FIG. 3 shows an example of the problem information stored in the problem information storage unit 232. As shown in the figure, the problem information has a structure in which problem groups are associated with each case.
In a case method course in which an instructor actually exists, for example, the instructor presents a certain case (subject), and then discussions and the like proceed based on that case.
One case data is information indicating the content of the case presented to the user in the case method course. The case data is, for example, in the form of text data, and has a content for explaining the case in a relatively long sentence. Specific examples of the contents of the cases shown by the case data are as follows.

"Airline A was founded in 19AA. Initially, it offered low prices and services comparable to flag carriers, and the annual number of passengers increased rapidly from 5,000 to 640,000 in the five years since its founding, but at the same time. , Recorded a large deficit from high-cost management, and was reported to be on the verge of breaking.
In 19BB, when he was forced to review his management policy, Mr. M, who is still the CEO (Chief Executive Officer), joined the management team. Airline A has relaunched as Europe's first low-cost carrier, modeled after US Airlines B, which has made great strides in low-cost and low-cost management.
Mr. M's strategy to target holiday customers was successful. Although business performance was temporarily sluggish due to the Gulf War from 19CC, a few years later, the route was expanded not only to the country where it was based, but also to the surrounding countries, and both the number of passengers and operating income have grown to the present. continuing.
(Omitted)
Immediately after the 9/11 terrorist attacks in the United States, air traffic demand itself declined significantly, but the impact on airline A, whose routes are limited to Europe, is limited, and leisure travelers who want to travel within Europe I was able to capture it. In addition, business travelers who have no choice but to fly have begun to choose low-cost carriers that depart and arrive at small airports, which are less likely to be targeted, avoiding the complexity of security at large airports. "

The problem group data is data having a plurality of problems to be given to the user U under the case indicated by the corresponding case data.
In the case method course conducted by a real-life instructor, the instructor can proceed with the discussion as follows. In other words, for example, the instructor poses a certain problem to the students based on the case. When the student returns an answer to the question, the instructor will further improve and deepen the student's thinking according to the content of the returned answer. Try to raise another issue. In this case, the instructor does not decide the next question in advance, but decides what kind of question to ask next according to the content of the answer. In other words, under the case method course, it is necessary to prepare a plurality of contents as the contents of the next problem following one problem. The problem group data of this embodiment has a structure corresponding to such a request.

FIG. 4 shows a configuration example of problem group data corresponding to the case data of 1. As described above, the question group data in the figure has a structure in which the contents of the next question can be appropriately changed according to the contents of the answer to one question.
In the figure, one problem included in the problem group data is regarded as a primary problem. An example in which three feedbacks (feedback A, feedback B, and feedback C) are associated with each other is shown for the primary problem. The feedback corresponds to the comment from the instructor's point of view presented to the user U according to the content of the answer to the question. Such feedback is created according to the combination of the question and the content of the answer. The content of the feedback may be, for example, a comment that evaluates the content of the answer, a comment that can give the user U notice by receiving the content of the answer, or the like in a sentence format.

In the example of the figure, the content of the answer is classified into, for example, one of three evaluation categories 1 to 3 prepared in advance, depending on the analysis targeting the answer to the primary problem.
As a specific example, for example, when the primary question requires that a specific matter is mentioned as an answer, the answer determined to refer to the specific matter is evaluated as "evaluation classification". 1. Answers that are judged not to mention specific matters are classified into evaluation classification 2. In addition, in the evaluation classification of the answers in such an example, the classification result is the above evaluation classification 1. However, depending on the content of the answer, the probability of falling into either evaluation category 1 or evaluation category 2 is low, and it can be classified into either evaluation category 1 or evaluation category 2. It may be difficult to classify such answers. Such answers are classified into evaluation classification 3. That is, evaluation classification 3 is a classification when proper evaluation is not possible.

For example, feedback A, feedback B, and feedback C correspond to evaluation classification 1, evaluation classification 2, and evaluation classification 3, respectively. Feedback A, feedback B, and feedback C each have the content of feedback according to the corresponding evaluation classification. The feedback A corresponding to the evaluation classification of the evaluation classification 1 has a content intended to convey, for example, that the user is making a reference to a specific matter required by the problem, and to lead to a better idea. In addition, the feedback B corresponding to the evaluation classification of the evaluation classification 2 points out, for example, that the specific matter required by the problem cannot be mentioned, and the content intended to lead to a more accurate idea. Have. Further, the feedback C corresponding to the evaluation classification of the evaluation classification 3 is, for example, the content based on a general comment or the like that does not correspond to either the feedback according to the evaluation classification of the evaluation classification 1 or the feedback according to the evaluation classification 2. It may have.

Further, one secondary problem is associated with each of the feedback A, the feedback B, and the feedback C. The secondary problem is the problem presented to the user after the primary problem. In the figure, the feedback A is associated with the secondary problem 1, the feedback B is associated with the secondary problem 2, and the feedback C is associated with the secondary problem 3.
That is, the content of the answer to the primary question is classified into any one of the three evaluation classifications of evaluation classification 1, evaluation classification 2, and evaluation classification 3. Then, according to the classification, the feedback for the primary question and the secondary question to be asked after the primary question are determined. Specifically, according to the fact that the answer to the primary question is classified into the evaluation classification 1, feedback A is selected as the feedback for the primary question, and the secondary question 1 is selected as the next question to the primary question.

Then, for each secondary question, feedback on the answer to the secondary question and a tertiary question to be given after the secondary question are determined according to the content of the answer. In the figure, as specific examples, feedback and a tertiary question according to the answer to the secondary question 2 are shown. That is, in the case of the secondary question 2, it is classified into one of the four evaluation categories 1 to 4 according to the content of the answer.
As a specific example, when the secondary problem 2 requires that a problem related to a certain matter be pointed out and a solution be derived based on the pointed out problem, the following is given according to the answer. It is classified into evaluation classification 1 to evaluation classification 4. In other words, the answer that accurately points out the problem and accurately derives the solution is classified into evaluation classification 1. In addition, the answers that could be pointed out accurately but did not accurately derive the solution are classified into evaluation classification 2. In addition, the answer in which the problem cannot be pointed out accurately and the solution cannot be accurately derived for this reason is classified into the evaluation classification 3. In addition, the answers that are judged to have low accuracy corresponding to any of the evaluation categories 1 to 3 are classified into the evaluation category 4. As an example, as an answer classified into the evaluation classification 4, for example, there may be an answer in which the solution is accurate even though the problem has not been accurately pointed out. In addition, answers that cannot be properly evaluated because they include points out problems, solutions, etc. that are completely irrelevant may also be classified into evaluation category 4.
In the figure, the feedback D and the tertiary problem 1 are associated with the evaluation classification 1. That is, when the answer to the secondary question 2 is classified into the evaluation of the evaluation classification 1, the feedback D and the tertiary question are given as the feedback for the secondary question 2 and the tertiary question to be asked after the secondary question 2, respectively. Problem 1 is decided.
Further, the feedback E and the tertiary problem 2 are associated with the evaluation classification 2. Further, the feedback F and the tertiary problem 2 are associated with the evaluation classification 3. As described above, for the question to be asked next, one next question may be common to a plurality of classifications of the answers to the previous questions. Although not shown, one feedback may be common to a plurality of classifications of answers to problems.
Further, the feedback G and the tertiary problem 3 are associated with the evaluation classification 4.
As described above, the problem group data has a structure in which one feedback and the next problem are associated with each evaluation classification prepared for one problem.

In addition, based on the structure of the question group data in the figure, the last question does not have to be determined as the number of questions, and it depends on the path of the questions that have been sequentially traced from the first question. May be made. For example, when the answer of the user U is excellent as a whole, the target learning effect can be obtained with a small number of questions, so that the number of questions to be asked may be small. On the other hand, in the case of a user who cannot easily return a good answer, it is necessary to have many questions answered with twists and turns until the target learning effect is obtained. In consideration of this, as a result of constructing the problem in the problem group data, the number of the last problem may be different depending on the path of the problem given to the user U.
As the structure of the question group data in the figure, a set (combination) of feedback and the next question is determined for one answer. However, for one answer, feedback is first determined, and then multiple candidates for the next question are linked to the determined feedback, and depending on some conditions, one of the candidates is linked. It may have a structure in which the next problem is determined.

Further, according to the structure of the question group data of the present embodiment, the user U can receive feedback on the answer each time the question is answered. As a result, even if it is an online course, it is possible to proceed with learning as if it were actually discussed with the instructor in the classroom.

An example of a processing procedure executed by the user terminal device 100 and the learning support device 200 of the present embodiment will be described with reference to the flowchart of FIG.
First, an example of a processing procedure executed by the user terminal device 100 will be described.
Step S101: When starting the case method learning, the user U performs an operation (login operation) of logging in the user terminal device 100 to the website of the case method course provided by the learning support device 200. In response to the login operation, the user terminal device 100 accesses the learning support device 200 via the network NT and executes the login request.

Step S102: By logging in to the website of the case method course, a web page on which a case to be used for future case method learning can be selected is displayed on the display unit of the user terminal device 100, for example. The user U performs an operation of selecting a case to be learned from the displayed web page. In response to the operation of selecting a case, the learning support device 200 reads the case data corresponding to the selected case from the problem information storage unit 232 and transmits the read case data to the learning support device 200. ..
The user terminal device 100 displays the content (case) of the case data transmitted from the learning support device 200 on the display unit. As a result, the user U can read the displayed case and grasp the contents.

The learning support device 200 may select a case suitable for the characteristics of the user U based on the characteristics evaluated for the user U in advance.
Alternatively, for example, there is a case where a learning plan for a long-term case method is set up, and a case in which the user U should take a course is determined according to the progress of learning according to the plan. In such a case, the user terminal device 100 does not select an example. In this case, the learning support device 200 determines a case to be used in this course according to the progress of learning according to the user U's plan, and reads out the case data of the determined case from the problem information storage unit 232. The case data may be transmitted to the learning support device 200.

Step S103: After grasping the content of the case, the user U is made to perform, for example, an operation for instructing the start of the question (question start instruction operation). In response to the question start instruction operation, the user terminal device 100 transmits the question start instruction to the user terminal device 100. In the user terminal device 100, the trigger for transmitting the question start instruction is not limited to the question start instruction operation. For example, a question start instruction may be transmitted when a certain period of time has passed since the case was displayed.

Step S104: After transmitting the question start instruction in step S103, the user terminal device 100 determines whether or not the problem data has been received.

Step S105: In response to the transmission of the question start instruction in step S103, the learning support device 200 first transmits the problem data of the problem as the first question. More specifically, the learning support device 200 may be configured to transmit a web page for asking the first question to the user terminal device 100. A web page for asking the first question is created so that the content of the question data of the question as the first question is displayed, for example, in sentences.
When the user terminal device 100 determines in step S104 that the problem data transmitted from the learning support device 200 has been received, the user terminal device 100 causes the display unit to display the problem of the content indicated by the received problem data.

In this embodiment, the problem data is based on characters (text) and images. Then, the problem displayed on the user terminal device 100 presents the content of the problem to the user U in sentences by text or an image. The problem data based on the image may be a moving image or a still image having the content of presenting the problem to the user U. Further, the image of the content that presents the problem to the user U may have a cartoon-like appearance. In response to the case mentioned above as a concrete example, for example, think about what kind of people were "holiday guests" and what kind of transportation they usually used. An example is given in which a problem with the content such as "is presented by sentences or images." In any case, natural language by sentences and voices is used to present problems by letters and images in this way.

Step S106: The user U reads the displayed question, considers an answer, and then performs an operation of inputting the considered answer (answer input operation). The answer input operation may be, for example, inputting a sentence (that is, a character string) into an input form displayed on the display unit.
That is, the answer to the problem in the present embodiment is not, for example, selection of options, but is written by the user U freely considering a sentence.

When the instructor and the student actually practice the case method, the instructor verbally conveys the problem to the user U, and the user U also verbally states the answer that he / she thinks. Therefore, in the present embodiment, the question to be asked by the user U is presented in the form of a sentence, and the answer is also described by the user U as a sentence. In other words, natural language is actively used for presenting cases, questions, and answers. As a result, the user U can perform learning with a feeling very close to that of an actual case method.

Step S107: When the user U finishes inputting the answer, for example, the user U performs an operation of instructing the transmission of the answer. In response to this operation, the user terminal device 100 transmits the input answer data (answer data) to the learning support device 200. The answer data transmitted by the learning support device 200 may include text data indicating a character string input as a sentence.

Step S108: The learning support device 200 transmits feedback data according to the content of the answer indicated by the answer data transmitted in step S107. The user terminal device 100 causes the display unit to display the feedback of the content indicated by the received feedback data. In addition, the feedback data may include the answers of other users that are helpful to the user U. In this case, step S108 displays the answers of other users, for example, along with feedback.

Step S109: After transmitting the answer data in step S107, the user terminal device 100 determines whether or not the comprehensive evaluation result has been received from the learning support device 200. The comprehensive evaluation result represents the result of performing a comprehensive evaluation on a plurality of answers of the user U so far according to the completion of all the questions.
If the comprehensive evaluation result is not received, the process is returned to step S104. If it is determined in step S104 that the problem is not received, steps S105 to S107 are skipped and the process proceeds to step S109. That is, the user terminal device 100 waits for either the problem or the comprehensive evaluation result to be received in a state where the processes of steps S105 to S107 have not been executed since the start of the question was transmitted in step S103. do.

The learning support device 200 transmits the question data as the second question in response to the transmission of the answer to the first question corresponding to the first case as described above. In this case, the learning support device 200 determines that the problem data has been received in step S104 in response to the reception of the problem data as the second question, executes the processes of steps S105 to S108, and performs learning support. The answer is transmitted to the device 200 and the feedback for the answer is displayed. After that, the user terminal device 100 repeats the processes of steps S105 to S108 each time the problem data transmitted from the learning support device 200 is received.

Step S110: Then, at a certain stage, the learning support device 200 transmits the final question, and the user terminal device 100 transmits the answer corresponding to the received final question in step S107. The learning support device 200 performs a comprehensive evaluation in response to receiving an answer to the final question, and transmits the comprehensive evaluation result to the user terminal device 100. Upon receiving the transmitted comprehensive evaluation result, the user terminal device 100 shifts from step S109 to the process of step S110, and displays the received comprehensive evaluation result on the display unit. The comprehensive evaluation result is displayed as a sentence. That is, the learning support device 200 generates a comprehensive evaluation result in the form of sentences.

Next, an example of the processing procedure executed by the learning support device 200 will be described.
Step S201: In the learning support device 200, the control unit 202 executes a process in response to a login request from the user terminal device 100. That is, the control unit 202 executes account authentication by using the user account and password included in the login request transmitted from the user terminal device 100. The control unit 202 permits access to the website for case method learning when the authentication is established.

Step S202: After the login is established, the question unit 221 of the control unit 202 transmits the case data selected from the case data stored in the problem information storage unit 232 to the user terminal device 100.

In selecting the case data to be transmitted to the user terminal device 100, the question unit 221 may select according to the case selection operation performed by the user on the web page as described above.
Alternatively, the question unit 221 may select case data according to the progress of the learning plan of the user U. In this case, the questioning unit 221 may refer to the learning progress status in the learning plan of the user U included in the user information stored in the user information storage unit 231 and determine the case to be used in this course.
Alternatively, the question unit 221 may select case data based on the characteristics evaluated for the user U. In this case, the questioning unit 221 may be configured to select case data using a learning model in which the learning device is trained with the characteristics of the user as learning data.

Step S203: After transmitting the case data in step S202, the learning support device 200 transmits a question start instruction in step S103. Upon receiving the question start instruction, the learning support device 200 shifts to the stage of asking a question (question) associated with the transmitted case data. Therefore, the questioning unit 221 substitutes "1" as an initial value for the variable n corresponding to the questioning order of the questions.

Step S204: The question unit 221 transmits the question data of the nth question (nth question). Here, when the question data as the first question (n = 1) is transmitted in step S204, the question unit 221 asks the first question from the question group data associated with the case data transmitted in step S202. The problem data of the problem is read from the problem information storage unit 232. The question unit 221 transmits the read question data.

Step S205: In response to the problem data being transmitted in step S204, the user terminal device 100 transmits the answer data reflecting the content of the answer input by the user U in step S107. In the learning support device 200, the answer analysis unit 222 acquires the answer data transmitted from the user terminal device 100.

Step S206: The answer analysis unit 222 determines whether or not the answer data acquired in step S205 corresponds to the answer to the last question.

Step S207: When it is determined that the answer does not correspond to the last question, the answer analysis unit 222 shifts to the process for asking the next question. Therefore, the answer analysis unit 222 analyzes the answer by using the answer data acquired in step S205. Here, the answer acquired as the answer data is, for example, in the form of a sentence using text data. That is, the answer in this embodiment is in natural language. Therefore, the answer analysis unit 222 analyzes the answer using natural language processing.
As an example, the answer analysis unit 222 may divide the character string indicated by the text as the answer data into words by morphological analysis and perform distributed expression for each word as the analysis of the answer. Further, the answer analysis unit 222 may perform context analysis such as syntactic analysis and semantic analysis on the character string indicated by the answer data as the analysis of the answer.
The answer analysis unit 222 derives items such as the content of the answer, the number of characters, the appropriateness of the vocabulary to be used (for example, easy-to-understand), etc. from the distributed expression for each word obtained as described above and the results of various analyzes. You can do it. In addition, the answer analysis unit 222 determines the difficulty level of the question for the user U or the characteristics of the user U, depending on, for example, the time (answer time) required from the transmission (question) of the question to the acquisition of the answer. (For example, the character tends to be thoughtful, or conversely, the answer tends to be decided without thinking too much). The answer analysis unit 222 classifies the answers based on the matters derived as described above. Specifically, as the answer classification, the answer analysis unit 222 responds to the problem, for example, evaluation classification 1 to evaluation classification N (N is a natural number of 2 or more) as described in the explanation of FIG. It may be classified into one of the preset evaluation classifications.

Here, the answer analysis unit 222 may be made to score the answer in the analysis of the answer in step S207. The answer analysis unit 222 may set a scoring standard in advance in response to a question, for example, and perform scoring based on whether or not the words, meanings, etc. obtained by the analysis satisfy the scoring standard. .. In addition, the answer analysis unit 222 may score the answers based on the results of classifying the answers.

Step S208: The determination unit 223 determines the feedback to be presented to the user U and the next question to be asked next based on the classification result obtained by the analysis of the answer in step S207.
Specifically, when the answer to the primary problem of FIG. 4 is analyzed in step S207 and the classification result with the evaluation classification 1 is obtained, the determination unit 223 presents the feedback to the user U and the next problem is shown in FIG. It is determined that the feedback A in 4 and the secondary problem 1 are present.

In the present embodiment, for example, a learning model can be used for analyzing the answer in step S207 and determining the next problem in step S208.
The learning model for the processing of steps S207 and S208 is based on a Transformer such as Support Vector Machine, LSTM (Long Short-Term Memory) with attention, or BERT (Bidirectional Encoder Representations from Transformers). Etc. may be adopted. In order to output the classification result according to the content of the answer, the learning device of such a learning model is made to input the answer of various contents as learning data and perform learning.
The learning model for the processing of steps S207 and S208 is not limited to the LSTM having the above-mentioned attention introduced, and other algorithms may be adopted.

The feedback and the next question determined in step S208 do not have the property of notifying the user U of the correctness of the answer obtained in step S205, but have the following contents, for example. That is, the feedback and the next question are designed to give notice to the user U who has given the answer, for example, according to the content of the answer obtained in step S205, and guide the user U to derive a better answer. Have content. In addition, the difficulty level of the next question can also be set according to the content of the question and the answer.
As an example, the case where three evaluation classifications 1 to 3 are set for the answer to a certain problem will be given as an example. Evaluation classification 1 evaluates that the correct answer element is not included, evaluation classification 2 evaluates that the correct answer element is included, and evaluation classification 3 determines whether or not the correct answer element is included. It is an evaluation that there is.
In such a case, the feedback corresponding to the evaluation classification 1 does not point out to the user U that it is incorrect, but an explanation that gives a hint in the direction of the correct answer while being an expression that is close to the user U. Can be the content to be performed for the user U. Further, as the next problem corresponding to the evaluation classification 1, it is possible to supplement the point that the user U lacks understanding. In addition, corresponding to the evaluation classification 2, feedback of the content that encourages the user U to think more applied ideas, and the next problem of the content that induces the user U to come up with the applied idea. can do. In addition, in response to evaluation classification 3, feedback and correct answers that give a sense of conviction by receiving and answering in expressions that are close to user U, regardless of whether the answer is closer to evaluation classifications 1 and 2. It can be the next problem that gives a hint in the direction of.
Further, as another example, a case where four evaluation classifications 1 to 4 are set for the answer to a certain problem will be given as an example. Evaluation classification 1 is the evaluation that both items A and B are included among the two items A and B required for the answer, and evaluation classification 2 is the evaluation that only item A is included among the items A and B, and the evaluation classification. 3 is an evaluation that includes only item B among items A and B, and evaluation classification 4 is an evaluation that neither item A nor B is included. In this case, corresponding to the evaluation classification 1, the feedback of the content that encourages the next thinking can be further set as the next problem. In addition, corresponding to the evaluation classification 2, the feedback of the content that induces the person to be aware of the matter B can be used as the next problem. Corresponding to the evaluation classification 3, the feedback of the content that induces the item A to be noticed can be set as the next problem. Corresponding to the evaluation classification 4, it can be a feedback of the content that induces the items A and B to be noticed, and the next problem.

According to the structure of the question group data in FIG. 4, the content of the next question depends on the content of the answer to the previous nth question. The nth question depends on the content of the answer to the previous (n-1) question. From this, it can be said that the content of the next question also depends on the content of the answers from the first question to the nth question. Then, it can be said that the combination of the contents of the first to the nth questions differs depending on the answer of the user U to each question. That is, in the present embodiment, the questions that are appropriate for the user U can be sequentially set for each user U according to the content of the answer for each user U.

For example, depending on the question, feedback or the next question may be determined in step S208 without performing the answer analysis in step S207. In this case, as the process of determining the feedback or the next problem in step S208, a single feedback or the next problem that is uniquely set in advance corresponding to the problem may be specified.
As an example of a problem that does not perform answer analysis in this way, a selection problem can be mentioned. That is, in the case method learning of the present embodiment, for example, in the flow of questions that are sequentially asked, not only the question of the content for which the answer is requested in natural language but also the selection question may be given as appropriate.
In this case, the answer from user U is one of the choices in the choice question, not in natural language. In this case, feedback and the next question are associated with each choice of the selection question in advance. When an answer to the selection question is obtained, the process of step S207 is skipped, and the determination unit 223 receives the feedback associated with the option selected as the answer by the user U in step S208, and the next question is the user. Feedback to be presented to U, determined as the next question.

Further, in determining the feedback, the determination unit 223 may include the answer selected from the answers of other users U to the same question corresponding to the answer to be analyzed in the feedback. At this time, the determination unit 223 may select an answer determined to be useful for the user U from among the answers of the other user U based on the analysis result of the answer. The criteria for determining the presence or absence of usefulness may include, for example, having a content that is close to the answer of the user U, or conversely, having a content that deviates from the answer of the user U. good. In addition, the criterion for determining the presence or absence of usefulness may include that the evaluation is higher than the evaluation for the answer of the user U.

Step S210: The question unit 221 increments the variable n and then returns the process to step S204 to transmit the next nth question to the user terminal device 100.
In this way, in the present embodiment, questions are sequentially asked from the first question according to one case. Here, the content of the question to be asked is determined according to the result of analyzing the content of the answer given by the user U to each of the questions given before that. Alternatively, depending on the question to be asked before, the next question, which is uniquely set in advance for the question, is given. Then, as the content of the problem, for example, as described above, the user U is guided to derive a better answer. And the problem has the content described as a sentence.
As a result, the user U can face the problem with the feeling that the instructor asks the user U. Further, when the answer of another user U is added to the next question, the idea of another user U can be absorbed. Then, each time the questions are asked in sequence, the user U answers the questions, so that the user U can proceed with the case method learning in a state close to the environment in which the instructor and other students are actually present. ..

Step S211: Then, the learning support device 200 gives the final question at a certain stage while the questions are given in sequence as described above. In response to the fact that the answer data for the last question was acquired in step S205, the answer analysis unit 222 determines that the answer to the last question was obtained in step S206. In this case, the answer analysis unit 222 analyzes the last question. The analysis according to step S211 may be the same as the process of step S207, for example. In this case, since the final answer is targeted, the answer is analyzed, but it is not necessary to determine the next question according to the answer analysis result as in step S208 following step S207.

Step S212: After the analysis of the last answer in step S211, the answer analysis unit 222 comprehensively evaluates the user U's previous answers based on the analysis results of the respective answers from the first question to the last question. Perform (comprehensive evaluation). In this case, the answer analysis unit 222 may output (generate) the comprehensive evaluation result in the form of sentences as described above. Further, the answer analysis unit 222 may use a learning model for the comprehensive evaluation in step S212. As the learning model in this case, for example, an LSTM having attention introduced may be adopted.
Alternatively, as a simpler configuration, for example, the answer analysis unit 222 associates a comprehensive evaluation with each evaluation classification of the answer to the last question in the question group data, and corresponds to the evaluation classification result of the answer to the last question. The overall evaluation to be performed may be obtained from the problem group data.
Step S213: The questioning unit 221 transmits the comprehensive evaluation result obtained in step S212 to the user terminal device 100.

In the present embodiment, the feedback and the question presented and given by the user terminal device 100 may be given by, for example, a moving image or an audio. Similarly, with respect to the answer, the user U may input the answer by recording himself / herself who is answering or recording the voice of the answer by using the user terminal device 100. In this case, the user terminal device 100 transmits the video recording (including audio) or audio data obtained by recording or recording as answer data. The answer analysis unit 222 of the learning support device 200 is made to extract a text in natural language from the voice and analyze the answer. When the answer is a moving image, the answer analysis unit 222 may analyze the emotion of the user U based on the facial expression, the gesture, and the like of the user U in the moving image. Further, the answer analysis unit 222 may analyze the emotion of the user U based on the tone of the voice in which the user is speaking the answer, etc., based on the voice data. The answer analysis unit 222 may also include the analyzed emotions of the user U in the input data for classifying the answers.

Note that the exchange between the question and the answer in this embodiment may be realized by, for example, a chat format.

The function of the learning support device 200 of the present embodiment is distributed to a plurality of devices on a network, for example, and the devices can cooperate with each other to provide a case method course to the user U. You may.

A program for realizing the functions of the user terminal device 100 and the learning support device 200 described above is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read by the computer system and executed. By doing so, the above-mentioned user terminal device 100, learning support device 200, and the like may be processed. Here, "loading and executing a program recorded on a recording medium into a computer system" includes installing the program in the computer system. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. Further, the "computer system" may include a plurality of computer devices connected via a network including a communication line such as the Internet, WAN, LAN, and a dedicated line. Further, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, or a storage device such as a hard disk built in a computer system. As described above, the recording medium in which the program is stored may be a non-transient recording medium such as a CD-ROM. The recording medium also includes an internal or external recording medium that can be accessed from the distribution server to distribute the program. The code of the program stored in the recording medium of the distribution server may be different from the code of the program in a format that can be executed by the terminal device. That is, the format stored in the distribution server does not matter as long as it can be downloaded from the distribution server and installed in a form that can be executed by the terminal device. The program may be divided into a plurality of parts, downloaded at different timings, and then combined by the terminal device, or the distribution server for distributing each of the divided programs may be different. Furthermore, a "computer-readable recording medium" is a volatile memory (RAM) inside a computer system that serves as a server or client when a program is transmitted via a network, and holds the program for a certain period of time. It shall also include things. Further, the above program may be for realizing a part of the above-mentioned functions. Further, it may be a so-called difference file (difference program) that can realize the above-mentioned function in combination with a program already recorded in the computer system.

100 user terminal device, 200 learning support device, 201 communication unit, 202 control unit, 203 storage unit, 221 question unit, 222 answer analysis unit, 223 decision unit, 231 user information storage unit, 232 problem information storage unit

Claims (6)

1. Through communication with the learner terminal device via the network, the problem in natural language is set on the learner terminal device, and the feedback to the learner according to the learner's answer to the problem is described above. The question section to be presented on the learner terminal device,
   An answer analysis unit that analyzes the answer to the problem in natural language acquired from the learner terminal device via communication via a network.
   A learning support device including a feedback unit presented by the questioning unit and a decision unit for determining the next question to be set by the questioning unit based on the analysis result by the answer analysis unit.
2. The learning support according to claim 1, wherein the determination unit selects an answer from the answers of other users based on the analysis result by the answer analysis unit, and includes the answer of the selected other user in the feedback. Device.
3. The answer analysis unit outputs a comprehensive evaluation of a plurality of answers to the one case based on the analysis results of the answers for each of the first to last questions corresponding to the one case. 2. The learning support device according to 2.
4. In the question section, in addition to the question in the natural language, it is possible to ask a choice question in which the choice selected from a plurality of choices is the answer.
   The answer analysis unit does not analyze the answer to the selection question, and the determination unit provides the feedback and the next question previously associated with the choices as the answer with the feedback presented by the question unit. , The learning support device according to any one of claims 1 to 3, which is determined as the next question to be asked by the questioning section.
5. Through communication with the learner terminal device via the network, problems in natural language are set on the learner terminal device, and feedback to the learner according to the learner's answer to the problem is given as described above. Question steps to be presented on the learner terminal device,
   An answer analysis step for analyzing a natural language answer to the problem obtained from the learner terminal device via communication via a network, and an answer analysis step.
   A learning support method including a feedback provided by the questioning step and a decision step of determining the next question to be asked by the questioning step based on the analysis result of the answer analysis step.
6. A computer as a learning support device,
   Through communication with the learner terminal device via the network, problems in natural language are set on the learner terminal device, and feedback to the learner according to the learner's answer to the problem is given as described above. The question section to be presented on the learner terminal device,
   An answer analysis unit that analyzes the answer to the problem in natural language acquired from the learner terminal device via communication via a network.
   A program for functioning as a decision unit for determining the feedback presented by the question section and the next question to be asked by the question section based on the analysis result by the answer analysis section.

Images