JP6460455B2 - Database construction device, learning support system, database construction method, learning support method, and program - Google Patents

Description

  The present invention relates to a database construction device, a learning support system, a database construction method, a learning support method, and a program.

  Conventionally, an adaptive test method has been proposed in which the relationship between proficiency levels for each unit is modeled by a Bayesian network, and a unit problem that maximizes the amount of network information is presented (for example, see Non-Patent Documents 1 and 2). . According to the learning support system using the methods shown in these Non-Patent Documents 1 and 2, it is possible to give a question after considering the correct / wrong situation of each unit of the learner and the relationship between the units. It is possible to grasp the learner's proficiency status efficiently with few problems.

Masao Ueno, Hitoshi Onishi, Kazuo Shigeki “Development of Adaptive Tests Using Teacher's Domain Structure Knowledge”, Journal of Japan Society for Educational Technology 18 (1), 15-24,1994 Masaomi Ueno, “Structural Optimization Method in Networked Test Theory”, IEICE Tech. Bulletin, ET93-ET137, 1994

  In the learning support system using the methods shown in Non-Patent Documents 1 and 2, each unit is associated with a Bayesian network node, and as the number of units increases, the computational load in the inference processing by the Bayesian network Will increase. For this reason, as the number of units increases, the time required to determine the next question to be asked becomes longer, and the waiting time until the next question is given becomes longer. There was a risk of damage.

  Therefore, the present invention has been made in view of the above-described problems, and an object of the present invention is to reduce a calculation load when determining a problem to be asked next.

  The present invention proposes the following items in order to solve the above-described problems. In addition, in order to make an understanding easy, although the code | symbol corresponding to embodiment of this invention is attached | subjected and demonstrated, it is not limited to this.

  (1) The present invention is a database construction device for constructing an inference result database and a question order database, and performs an adaptive test simulation using a learning log indicating a test result of a learner, and the execution process of the simulation In addition to storing the Bayesian network inference results required in the above inference result database, the relationship between the correctness of the learner's answers to each question in the exam and the unit to be presented next is stored in the above question order database. It proposes a database construction device characterized by comprising database construction means (for example, corresponding to the database construction unit 10 in FIG. 1).

  According to the present invention, the database construction means performs an adaptive test simulation using the learning log indicating the learner's test result, and the Bayesian network inference result required in the execution process of the simulation is obtained as the inference result database. In addition, the relationship between the correctness of the learner's answer to each question in this exam and the unit to be presented next is stored in the question order database. For this reason, referring to at least one of the inference result database and the question order database, it is possible to determine the unit of the problem to be asked next without performing inference processing by the Bayesian network. In addition, the inference result database and the question order database can be constructed in advance. Therefore, it is not necessary to perform an inference process using a Bayesian network when determining a unit of a problem to be asked next, so that the calculation load can be reduced.

  (2) In the database construction device according to (1), the database construction unit refers to at least one of the inference result database and the question order database when the simulation is performed. A database construction device is proposed.

  According to the present invention, in the database construction device of (1), at least one of the inference result database and the question order database is referred to by the database construction means when performing the adaptive test simulation. For this reason, when constructing the inference result database and the question order database, the frequency of performing the inference processing by the Bayesian network can be reduced, so that the calculation load can be reduced.

  (3) The present invention is a learning support system (for example, corresponding to the learning support system 1 of FIG. 1) including the inference result database and the question order database constructed by the database construction device of (1) or (2). Then, referring to at least one of the inference result database and the question order database, an adaptive test operation means (for example, corresponding to the adaptive test operation unit 30 in FIG. And the adaptive test operation means uses the Bayesian network to determine the unit when the next unit to be presented cannot be determined by referring to either the inference result database or the question order database. We propose a learning support system characterized by decision.

  According to the present invention, in the learning support system including the inference result database and the question order database constructed by the database construction device of (1) or (2), the inference result database and the question order database are adapted by the adaptive test operation means. With reference to at least one of them, we decided to decide the unit to be asked next. For this reason, it is not necessary to perform inference processing by the Bayesian network when determining the unit of the problem to be asked next, so that the calculation load can be reduced.

  According to the present invention, in the learning support system including the inference result database and the question order database constructed by the database construction device according to (1) or (2), the inference result database and the question order are arranged by the adaptive test operation means. If no unit to be examined next can be determined by referring to any of the databases, the unit is determined using the Bayesian network. For this reason, if you cannot determine the unit that should be given the next question by referring to either the inference result database or the question order database, perform inference processing using the Bayesian network to determine the unit that should be given the next question. Can do.

  (4) In the learning support system according to (3), when the adaptive test operation unit determines the unit using the Bayesian network, the inference result using the Bayesian network is used as the inference result. Proposing a learning support system characterized in that it stores at least one of storing in a database and storing the relationship between correctness of the answer and the determined unit in the question order database. .

  According to the present invention, in the learning support system of (3), when the unit is determined using the Bayesian network by the adaptive test operation means, the inference result using the Bayesian network is stored in the inference result database; The relationship between the correctness of the answer and the determined unit is stored in the question order database, and at least one of them is performed. For this reason, when the unit to be given next question is determined, the inference result database and the question order database can be appropriately updated based on the result obtained at the time of the determination.

  (5) The present invention is a database construction method in a database construction device that comprises database construction means (for example, equivalent to the database construction unit 10 in FIG. 1) and constructs an inference result database and a question order database, and the database construction The means performs a simulation of an adaptive test using a learning log indicating the test result of the learner, stores the inference result of the Bayesian network required in the execution process of the simulation in the inference result database, and the test Has proposed a database construction method characterized by comprising a first step of storing a relationship between correctness of a learner's answer to each question and a unit to be presented next in the question order database.

  According to the present invention, in the first step, a simulation of an adaptive test using a learning log indicating a learner's test result is performed, and an inference result of a Bayesian network necessary in the process of executing the simulation is inferred. In addition to being stored in the database, the relationship between the correctness of the learner's answers to each question in this exam and the unit to be asked next was stored in the question order database. For this reason, referring to at least one of the inference result database and the question order database, it is possible to determine the unit of the problem to be asked next without performing inference processing by the Bayesian network. In addition, the inference result database and the question order database can be constructed in advance. Therefore, it is not necessary to perform an inference process using a Bayesian network when determining a unit of a problem to be asked next, so that the calculation load can be reduced.

  (6) The present invention includes adaptive test operation means (for example, equivalent to the adaptive test operation unit 30 of FIG. 1), and the inference result database and the question order database constructed by the database construction method of (5) A learning support system (for example, equivalent to the learning support system 1 in FIG. 1), wherein the adaptive test operation means refers to at least one of the inference result database and the question order database. The second step for determining the next unit to be presented, and the adaptive test operation means determines the unit to be presented next regardless of which of the inference result database and the question order database. A third step of determining the unit using a Bayesian network if it is not possible. It has proposed a learning support method.

  According to the present invention, in the learning support method in the learning support system including the inference result database and the question order database constructed by the database construction method of (5), the second step includes the inference result database and the question order database. By referring to at least one of them, we decided to decide the unit to be asked next. For this reason, it is not necessary to perform inference processing by the Bayesian network when determining the unit of the problem to be asked next, so that the calculation load can be reduced.

  According to the present invention, in the learning support method in the learning support system comprising the inference result database and the question order database constructed by the database construction method of (5), the inference result database and the question order database are provided in the third step. If the unit to be asked next cannot be determined by referring to any of these, the unit is determined using the Bayesian network. For this reason, if you cannot determine the unit that should be given the next question by referring to either the inference result database or the question order database, perform inference processing using the Bayesian network to determine the unit that should be given the next question. Can do.

  (7) The present invention comprises a database construction method in a database construction apparatus that comprises database construction means (e.g., equivalent to the database construction unit 10 in FIG. 1) and constructs an inference result database and a question order database. The database construction means performs a simulation of an adaptive test using a learning log indicating a learner's test result, and the inference result of a Bayesian network required in the execution process of the simulation A first step of storing in the database the first step of storing in the question order database the relationship between the correctness of the learner's answers to each question in the examination and the unit to be given next Propose a program.

  According to the present invention, by executing a program using a computer, the first step performs a simulation of an adaptive test using a learning log indicating a test result of a learner. Store the necessary Bayesian network inference results in the inference result database, and store the relationship between the correctness of the learner's answers to each question in this exam and the unit to be presented next in the question order database. did. For this reason, referring to at least one of the inference result database and the question order database, it is possible to determine the unit of the problem to be asked next without performing inference processing by the Bayesian network. In addition, the inference result database and the question order database can be constructed in advance. Therefore, it is not necessary to perform an inference process using a Bayesian network when determining a unit of a problem to be asked next, so that the calculation load can be reduced.

  (8) The present invention includes adaptive test operation means (for example, equivalent to the adaptive test operation unit 30 in FIG. 1), and the inference result constructed by executing the program of (7) using a computer. A program for causing a computer to execute a learning support method in a learning support system (e.g., corresponding to the learning support system 1 in FIG. 1) including a database and the question order database, wherein the adaptive test operation means includes the A second step of determining a unit to be given next by referring to at least one of the inference result database and the question order database; and the adaptive test operation means includes the inference result database and the question order database If you cannot determine which unit to ask next, We propose a program to execute a third step of determining the Unit, to the computer using the Tsu bets.

  According to the present invention, in the learning support method in the learning support system including the inference result database and the question order database constructed by executing the program of (7) using a computer, the program is executed using the computer. In the second step, the next unit to be asked is determined with reference to at least one of the inference result database and the question order database. For this reason, it is not necessary to perform inference processing by the Bayesian network when determining the unit of the problem to be asked next, so that the calculation load can be reduced.

  According to the invention, in the learning support method in the learning support system including the inference result database and the question order database constructed by executing the program of (7) using a computer, the program is executed using the computer. Therefore, in the third step, when it is not possible to determine the next unit to be asked by referring to either the inference result database or the question order database, the unit is determined using the Bayesian network. . For this reason, if you cannot determine the unit that should be given the next question by referring to either the inference result database or the question order database, perform inference processing using the Bayesian network to determine the unit that should be given the next question. Can do.

  According to the present invention, it is possible to reduce a calculation load when determining a problem to be asked next.

1 is a block diagram of a learning support system 1 according to an embodiment of the present invention. It is a figure which shows an example of the inference result database which concerns on one Embodiment of this invention. It is a figure which shows an example of the question order database which concerns on one Embodiment of this invention. It is a flowchart which shows the process which the database construction part concerning one Embodiment of this invention performs. It is a flowchart which shows the process which the database construction part concerning one Embodiment of this invention performs. It is a flowchart which shows the process which the adaptive type test operation part which concerns on one Embodiment of this invention performs.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. Note that the constituent elements in the following embodiments can be appropriately replaced with existing constituent elements, and various variations including combinations with other existing constituent elements are possible. Accordingly, the description of the following embodiments does not limit the contents of the invention described in the claims.

  FIG. 1 is a block diagram of a learning support system 1 according to an embodiment of the present invention. The learning support system 1 includes a database construction unit 10 that constructs an inference result database and a question order database in advance, a storage unit 20 that stores an inference result database and a question order database constructed by the database construction unit 10, and a storage unit 20. And an adaptive test operation unit 30 that asks the learner questions by referring to the stored inference result database and the question order database.

[Inference result database]
The inference result database stores the inference results of the Bayesian network that are necessary in the process of executing the adaptive test simulation. With reference to this inference result database, the probabilities of each unit for which test results have not yet been obtained can be obtained as search results using the proficiency status of units for which test results have already been obtained as search keys.

  FIG. 2 is a diagram illustrating an example of an inference result database. In FIG. 2, “−1” indicates an unanswered unit, “0” indicates an unskilled unit, and “1” indicates a mastered unit. The units U1 to Um correspond to, for example, mathematics, such as addition, multiplication, integration, etc. (m is an integer satisfying m> 1).

  As can be seen from FIG. 2, the units U2 and Um are unanswered. This means that the units U2 and Um are units for which test results have not yet been obtained.

  The learning probabilities of the units U2 and Um are inferred by the database construction unit 10 and stored in the inference result database. The probabilities of proficiency are probabilities of proficiency and probabilities of no proficiency, and the sum of probabilities of proficiency and probabilities of no proficiency is 1.0. Therefore, if one of the probabilities of proficiency and the probability of not proficient is known, the other probability can be obtained. Therefore, in the case of the proficiency status of units U1 to Um shown in FIG. 2, the probability that the unit U2 is proficient is 0.8, and the probability that the unit U2 is not proficient is 0.2. At least one of the information that exists is stored in the inference result database. Further, in the case of the proficiency status of units U1 to Um shown in FIG. 2, the probability that the unit Um is proficient is 0.4, and the probability that the unit Um is not proficient is 0.6. At least one of the information that exists is stored in the inference result database.

  An example of the above reasoning will be described below.

  For example, knowledge of addition is essential for performing multiplication. For this reason, if the unit of addition is not mastered, it can be assumed that the unit of multiplication is not mastered. Therefore, when the unit of addition is unfamiliar and the unit of multiplication is unanswered, it can be inferred that the probability that the unit of multiplication is not mastered is lower than the probability of mastering the unit of multiplication.

  On the other hand, if you are familiar with the unit of multiplication, you can assume that you are also familiar with the unit of addition. Therefore, if the multiplication unit is already mastered and the addition unit is unanswered, it can be inferred that the probability of mastering the addition unit is higher than the probability of not mastering the addition unit. .

  As described above, there is a relationship between units such that if one unit is mastered, the other unit will be mastered. Therefore, this relationship is used in the above reasoning. Note that the network structure representing this relationship and the conditional probability between the related units can be determined by the administrator of the learning support system 1 or the teaching material creator. In addition, if there is a result of actual test, it can be obtained by automatically learning the network structure and conditional probability by processing this test result using Bayesian network related technology. .

  Note that the learning log R of n learners for m types of units can be expressed by the following formula (1) (n is an integer satisfying n ≧ 1). The learning log is information indicating past test results by a plurality of learners, and each question presented in the test is linked to at least one of a plurality of units. Assume that more than one problem is linked. Each unit is associated with a node in the Bayesian network, and the state of each node is associated with the answer status (correct / incorrect information) for the problem associated with the unit associated with that node. . It is assumed that the learning log has information on whether or not each learner has correctly answered all the problems except when there is a missing value for a special reason.

In Equation (1), r _{i, j} represents the learning status of the learner i regarding the unit Uj (i is an arbitrary integer satisfying 1 ≦ i ≦ n, and j satisfies 1 <j ≦ m. Meet any integer). In the case where there is one problem associated with the unit Uj, “1” is entered in r _{i, j} if the learner i has correctly answered this problem. If so, “0” is entered. In addition, when there are a plurality of problems associated with the unit Uj, the ratio of the correct answers of the learner i out of the plurality of problems is totaled. If this ratio is equal to or greater than the threshold, “1” is entered. If this ratio is less than the threshold, “0” is entered. In addition, when there are a plurality of problems associated with the unit Uj, the ratio of the correct answers of the learner i out of the plurality of problems is counted, and the ratio is quantized according to a predetermined rule. (For example, “0” if the accuracy rate is 0 or more and less than 0.25, “1” if the accuracy rate is 0.25 or more and less than 0.5, and the accuracy rate is 0.00). “2” if it is 5 or more and less than 0.75, and “3” if the accuracy rate is 0.75 or more and 1.0 or less).

[Question order database]
The question order database stores the relationship between correct and incorrect information of the learner's answers to each question and the unit to be given next. By referring to this question order database, the unit to be presented next can be obtained as a search result using the proficiency of the unit for which the test result has already been obtained as a search key.

  FIG. 3 is a diagram illustrating an example of a question order database. In FIG. 3, as in FIG. 2, “−1” indicates an unanswered unit, “0” indicates an unskilled unit, and “1” indicates a mastered unit.

  According to FIG. 3, it can be seen that the units U2 and Um are unanswered. This means that the units U2 and Um are units for which test results have not yet been obtained.

  The unit to be given next among the units U2 and Um is inferred by the database construction unit 10 and stored in the question order database. In the case of the proficiency status of units U1 to Um shown in FIG. 3, the information that the unit to be given next is unit U2 is stored in the question order database.

  An example of determining the unit to be presented next will be described below.

  For example, it is assumed that an addition unit, a multiplication unit, and an integration unit are unanswered. In this case, the network-type item information amount shown in “Masao Ueno, Hitoshi Onishi, Kazuo Shigeru“ Proposal of Test Theory with Stochastic Network ”, IEICE Technical Report, ET93-64, 55-62,1993” EVINI (Expected Value of Item Network Information) is calculated for the above three unanswered units, and it is determined that the unit with the largest EVINI, that is, the unit with the most information should be given next.

[Operation of Database Building Unit 10]
Returning to FIG. 1, the database construction unit 10 constructs the inference result database and the question order database in advance as described above. Details of the database construction unit 10 will be described below.

  First, the database construction unit 10 learns the Bayesian network model and the conditional probability table using the learning log R.

  Next, the database construction unit 10 performs an adaptive test simulation using the learning log used when learning the Bayesian network model, and constructs the inference result database and the question order database.

  4 and 5 are flowcharts showing processing performed by the database construction unit 10.

  In step S1, the database construction unit 10 substitutes 1 for i, and moves the process to step S2. According to this, the first learner is selected from the n learners described above.

  In step S2, the database construction unit 10 determines whether i is n or less. If it is determined that the number is less than or equal to n, the process proceeds to step S3. If it is determined that the number is greater than n, the process illustrated in FIGS.

In step S3, the database construction unit 10 substitutes the learning log R _i of the i-th learner into the array LIST_1, initializes the array LIST_2, and sets all elements of the array LIST_2 to “−1”. Move processing to. The array LIST_1 is represented by the following formula (2), and the array LIST_2 is represented by the following formula (3).

  In step S4, the database construction unit 10 refers to the question order database using the array LIST_2 as a key, and when the learning status of the units U1 to Um is represented by the array LIST_2 in the question order database, the next question is given. It is determined whether or not there is information about the power unit. If it is determined that there is, the process proceeds to step S5. If it is determined that there is not, the process proceeds to step S6.

  In step S5, the database construction unit 10 refers to the question order database using the array LIST_2 as a key, and if the learning status of the units U1 to Um is represented by the array LIST_2, A determination is made from U1 to Um, and the process proceeds to step S12.

  In step S <b> 6, the database construction unit 10 refers to the inference result database and determines whether it is possible to determine a unit to be given next. Specifically, in the inference result database, it is determined whether or not the probabilities of each unit for which test results have not yet been obtained when the proficiency status of units U1 to Um is represented by the array LIST_2 To do. If it is determined that there is, the process proceeds to step S7, and if it is determined that there is not, the process proceeds to step S9.

  In step S7, the database constructing unit 10 determines a unit j to be presented next by EVINI based on the value of the array LIST_2, and moves the process to step S8. When calculating EVINI, the search result stored in the inference result database is used as the proficiency level inference result for the unanswered unit.

  In step S8, the database construction unit 10 associates the unit j determined to be the next question in step S7 and the array LIST_2, stores them in the question order database, and moves the process to step S12.

  In step S9, the database construction unit 10 performs proficiency level inference using a Bayesian network, determines a unit j to be presented next, and moves the process to step S10.

  In step S10, the database construction unit 10 acquires the proficiency pattern (the proficiency level of the unit) input as evidence (determined value) at the time of inference by the Bayesian network in step S9, and the inferred proficiency level (test results have not yet been obtained) The relationship between the learning probabilities of each unit) is stored in the inference result database, and the process proceeds to step S11.

  In step S11, the database construction unit 10 stores the unit j determined to be the next question in step S9 and the array LIST_2 in the question order database, and moves the process to step S12. When the array LIST_2 satisfies the end condition, that is, when the questions are to be ended, “0” is stored in the question order database.

In step S12, the database structuring unit 10, element list_2 with substituting elements list_1 _j to _j, indicating that already questions to the element list_1 _j substitutes "-999", the process proceeds to step S13.

  In step S <b> 13, the database construction unit 10 satisfies both that the element other than “−999” exists in the array LIST_1 and that the question unit is a value other than “0”. The process returns to step S4. If at least one of the conditions is not satisfied, the process proceeds to step S14.

  In step S14, the database construction unit 10 adds “1” to i, and returns the process to step S2.

[Operation of Adaptive Test Operation Unit 30]
Returning to FIG. 1, as described above, the adaptive test operation unit 30 refers to the inference result database and the question order database stored in the storage unit 20 and presents the question to the learner. Details of the adaptive test operation unit 30 will be described below.

  The adaptive test operation unit 30 refers to the question order database and uses the current proficiency level of the learner asking the question as a query to determine a unit to be given next for this learner.

  In addition, when the unit that should be given the next question cannot be determined by referring to the question order database, the adaptive test operation unit 30 determines and decides the unit that should be given next by the adaptive testing mechanism. The unit and the current proficiency status of the learner who questions the question are linked and stored in the question order database. When calculating EVINI in the adaptive testing mechanism, the inference result database is referred to.

  Further, if the adaptive test operation unit 30 cannot search for the unit to be asked next even if referring to the question order database, and there is no value in the inference result database, first, learning using the Bayesian network is performed. Perform inference and determine the next unit to be presented. Next, the inference result of the proficiency level is stored in the inference result database, and the determined unit and the current proficiency status of the learner who questions the question are linked and stored in the question order database. In the case where there is no value in the inference result database, since the reasoning for the current proficiency of the learner who has asked the question has not been made so far, the value (probability of proficiency) for the search key (proficiency) is That is, it is not in the inference result database.

  FIG. 6 is a flowchart showing processing performed by the adaptive test operation unit 30.

  In step S21, the adaptive test operation unit 30 initializes the array LIST, and moves the process to step S22. Note that the array LIST is represented by the following mathematical formula (4).

As a result of the initialization in step S21, each element list _p (p is an arbitrary integer satisfying 1 ≦ p ≦ m) of the array LIST in Expression (4) is all “−1”. The value of the element list _p is updated according to the correct / incorrect result of the learner's answer to the question that has been given. Specifically, when the unit up question is given in step S29, which will be described later, if the learner's answer is correct, the value of the element list _p is updated to “1”, and the learner's answer is incorrect. If there is, the value of the element list _p is updated to “0”. For example, if the unit U1 is already mastered among the five units, the unit U2 is unskilled, and the units U3, U4, and U5 are unanswered, the array LIST is expressed by the following equation (5). Therefore, by referring to the question order database using each element list _p of the array LIST as a search key, a unit to be given next can be obtained.

  In step S22, the adaptive test operation unit 30 refers to the question order database using the array LIST as a key, and if the learning status of the units U1 to Um is represented by the array LIST in the question order database, It is determined whether there is information about a unit to be given. If it is determined that there is, the process proceeds to step S23, and if it is determined that there is not, the process proceeds to step S24.

  In step S23, the adaptive test operation unit 30 first refers to the question order database using the array LIST as a key, and if the proficiency status of the units U1 to Um is represented by the array LIST, the next question should be given. A unit is determined from the units U1 to Um. Next, a value indicating the unit determined to be the next question among 1 to m is substituted into the variable U, and the process proceeds to step S29. For example, if the array LIST is in the state shown in FIG. 3, the unit to be presented next is determined to be the unit U2, and 2 is substituted into the variable U.

  In step S <b> 24, the adaptive test operation unit 30 refers to the inference result database and determines whether the next unit to be presented can be determined. Specifically, in the inference result database, it is determined whether or not there is a learning probability of each unit for which the test result has not yet been obtained when the learning status of the units U1 to Um is represented by the array LIST To do. If it is determined that there is, the process proceeds to step S25, and if it is determined that there is not, the process proceeds to step S26.

  In step S25, the adaptive test operation unit 30 refers to the inference result database, determines a unit to be presented next by the adaptive testing mechanism, and moves the process to step S29. For example, when the array LIST is in the state shown in FIG. 2, it is determined that the next unit to be presented is the unit U2.

  In step S26, the adaptive test operation unit 30 performs proficiency level inference using a Bayesian network, determines a unit to be presented next, and moves the process to step S27.

  In step S27, the adaptive test operation unit 30 stores the relationship between the array LIST and the proficiency level inference result in step S26 in the inference result database, and the process proceeds to step S28.

  In step S28, the adaptive test operation unit 30 stores the relationship between the array LIST and the unit determined to be the next question in step S26 in the question order database, and the process proceeds to step S29.

In step S29, the adaptive test operation unit 30 asks the learner the question of the unit Uj determined to be the next question, and substitutes the correct / incorrect result of the learner's answer to the LIST _j for the question given. Then, the process proceeds to step S30.

  In step S30, the adaptive test operation unit 30 updates the array LIST according to the correct / incorrect result of the learner's answer to the question presented in step S29, and the process proceeds to step S31.

  In step S <b> 31, the adaptive test operation unit 30 determines whether all the questions have been asked and whether the test has ended. If all questions have been given or it is determined that the test has been completed, the process proceeds to step S32. If all the questions have not been asked yet and it is determined that the test has not ended, the process returns to step S22.

  In step S32, the adaptive test operation unit 30 stores the array LIST that holds the answer status of the learner so far and “0” indicating the completion of the test in the question order database, as shown in FIG. The indicated process ends.

  According to the learning support system 1 described above, the following effects can be achieved.

  In the learning support system 1, the database construction unit 10 performs an adaptive test simulation using a learning log indicating a learner's test result, and an inference result of a Bayesian network inference result required in the course of the simulation is obtained. In addition to storing in the database, the relationship between the correctness / learning of the learner's answers (proficiency) for each question in this exam and the unit to be presented next is stored in the question order database. For this reason, referring to at least one of the inference result database and the question order database, it is possible to determine the unit of the problem to be asked next without performing inference processing by the Bayesian network. In addition, the inference result database and the question order database can be constructed in advance. Therefore, it is not necessary to perform an inference process using a Bayesian network when determining a unit of a problem to be asked next, so that the calculation load can be reduced.

  In the learning support system 1, the database construction unit 10 refers to at least one of the inference result database and the question order database when the adaptive test simulation is performed. For this reason, when constructing the inference result database and the question order database, the frequency of performing the inference processing by the Bayesian network can be reduced, so that the calculation load can be reduced.

  In the learning support system 1, the adaptive test operation unit 30 refers to at least one of the inference result database and the question order database, and determines a unit to be given next. For this reason, it is not necessary to perform inference processing by the Bayesian network when determining the unit of the problem to be asked next, so that the calculation load can be reduced.

  In addition, the learning support system 1 uses the Bayesian network when the adaptive test operation unit 30 cannot determine a unit to be given next, regardless of which one of the inference result database and the question order database is referred to. To decide. For this reason, if you cannot determine the unit that should be given the next question by referring to either the inference result database or the question order database, perform inference processing using the Bayesian network to determine the unit that should be given the next question. Can do.

  In addition, the learning support system 1 stores the inference result using the Bayesian network in the inference result database and determines whether the answer is correct or not when the adaptive test operation unit 30 determines the unit using the Bayesian network. Storing the relationship with the unit in the question order database. For this reason, when the unit to be given next question is determined, the inference result database and the question order database can be appropriately updated based on the result obtained at the time of the determination.

  The processing of the learning support system 1 according to the present invention is recorded on a computer-readable non-transitory recording medium, and the program recorded on the recording medium is read into the learning support system 1 and executed. The invention can be realized.

  Here, for example, a nonvolatile memory such as an EPROM or a flash memory, a magnetic disk such as a hard disk, a CD-ROM, or the like can be applied to the above-described recording medium. Further, reading and execution of the program recorded on the recording medium is performed by a processor provided in the learning support system 1.

  The above-described program may be transmitted from the learning support system 1 storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

  Further, the above-described program may be for realizing a part of the above-described function. Furthermore, what can implement | achieve the above-mentioned function in combination with the program already recorded on the learning assistance system 1, what is called a difference file (difference program) may be sufficient.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes a design that does not depart from the gist of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Learning support system 10 ... Database construction part 20 ... Memory | storage part 30 ... Adaptive test operation part

Claims (8)

A database construction device for constructing an inference result database and a question order database,
The database construction device
Performs an adaptive test simulation using a learning log indicating the learner's test results, stores the Bayesian network inference results required in the simulation execution process in the inference result database, and each problem in the test A database construction device comprising: database construction means for storing a relationship between correctness of a learner's answer to and a unit to be presented next in the question order database.   The database construction apparatus according to claim 1, wherein the database construction unit refers to at least one of the inference result database and the question order database when the simulation is performed. A learning support system comprising the inference result database constructed by the database construction apparatus according to claim 1 or 2 and the question order database,
An adaptive test operation means that determines at least one unit to be asked next by referring to at least one of the inference result database and the question order database;
The adaptive test operation means includes:
A learning support system, which determines a unit using a Bayesian network when a unit to be given next cannot be determined by referring to either the inference result database or the question order database.   When the unit is determined using the Bayesian network, the adaptive test operation means stores an inference result using the Bayesian network in the inference result database, whether the answer is correct, and the determined unit. 4. The learning support system according to claim 3, wherein at least one of storing the relationship with the question order database is performed. 5. A database construction method in a database construction device comprising a database construction means and constructing an inference result database and a question order database,
The database construction means of the database construction device performs a simulation of an adaptive test using a learning log indicating a test result of a learner, and obtains an inference result of a Bayesian network required in the execution process of the simulation. A database comprising a first step of storing in a database the relationship between correctness of learner's answers to each question in the examination and a unit to be given next, in the question order database Construction method. A learning support method in a learning support system comprising adaptive test operation means, comprising the inference result database constructed by the database construction method according to claim 5 and the question order database,
A second step in which the adaptive test operation means of the learning support system refers to at least one of the inference result database and the question order database, and determines a unit to be asked next;
If the adaptive test operation means of the learning support system cannot determine a unit to be asked next regardless of which of the inference result database and the question order database, the unit is determined using a Bayesian network. And a third step of determining the learning support method. A program for causing a computer to execute a database construction method in a database construction device that comprises a database construction means and constructs an inference result database and a question order database,
The database construction means of the database construction device performs a simulation of an adaptive test using a learning log indicating a test result of a learner, and obtains an inference result of a Bayesian network required in the execution process of the simulation. A first step of storing in the database the first step of storing in the question order database the relationship between the correctness of the learner's answers to each question in the examination and the unit to be given next program. A learning support method in a learning support system, comprising adaptive test operation means, comprising the inference result database constructed by executing the program according to claim 7 using a computer and the question order database, executed on a computer A program for
A second step in which the adaptive test operation means of the learning support system refers to at least one of the inference result database and the question order database, and determines a unit to be asked next;
If the adaptive test operation means of the learning support system cannot determine a unit to be asked next regardless of which of the inference result database and the question order database, the unit is used by using a Bayesian network. And a third step of determining a program for causing a computer to execute.

Images