JP7052224B2 - Learning support device, learning support system and program - Google Patents

Description

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

A method is known in which the results of tests conducted on learners such as students are analyzed based on Item Response Theory (IRT) to measure the degree of understanding of the learners. In the following description, the item response theory will be referred to as IRT for the sake of brevity.

For example, Patent Document 1 is a system for performing a computerized adaptive test using IRT, and performs IRT analysis on the test results of each learner to estimate and estimate the learner's comprehension level (scholastic ability). A system is disclosed that presents learners with difficulty problems according to their comprehension level. According to Patent Document 1, the reliability coefficient of the estimated comprehension is calculated, and it is determined that sufficient analysis accuracy is guaranteed when the reliability coefficient becomes a certain threshold value or more, and a new problem is presented.

Japanese Unexamined Patent Publication No. 2005-215023

However, in the invention according to Patent Document 1, although the reliability of the analysis result is evaluated, the validity of the test problem itself is not evaluated. Therefore, the learner may be made to learn a test question that is not valid as an evaluation scale for measuring the learner's comprehension, that is, a poor test question.

The present invention has been made under such circumstances, and an object of the present invention is to provide a learning support device or the like capable of supporting efficient learning.

The learning support device according to the present invention has a collection unit that collects answers to test items by a plurality of learners, a storage unit that stores the collected answers of the plurality of learners, and answers of the plurality of learners. Based on this, the analysis unit that analyzes the discriminating power of the problem from the item characteristics according to the learner's understanding level and the correct answer rate of the test item, and the problem that satisfies the predetermined condition according to the discriminating power in the analyzed item characteristics. It is characterized by having a transmitter for transmitting the above as a review question.

The learning support device according to the present invention includes a second storage unit that stores the review question corresponding to the test item, and an extraction unit that extracts the review question according to the item characteristics analyzed by the analysis unit. The transmission unit is characterized in that the review question extracted by the extraction unit is transmitted.

In the learning support device according to the present invention, the storage unit stores the answers of the plurality of learners in association with the answer date and time when each of the learners gave an answer, and the correctness of the answer of each of the learners. And, based on the elapsed time from the answer date and time, each learner is provided with an estimation unit that estimates the current fixation degree of the test item, and the transmission unit transmits the review question at a timing according to the fixation degree. It is characterized by doing.

The learning support device according to the present invention is characterized in that the estimation unit estimates the degree of fixation according to the attributes of the learner.

The learning support device according to the present invention is characterized in that the estimation unit estimates the degree of fixation according to the attributes of the test items.

The learning support device according to the present invention is characterized in that the estimation unit estimates the degree of fixation according to the item characteristics analyzed by the analysis unit.

The learning support system according to the present invention is a learning support system having a learning support device and a terminal device that communicates with the learning support device, and the learning support device is a test by a plurality of learners from the terminal device. Based on the collection unit that collects the answers to the items, the storage unit that stores the collected answers of the plurality of learners, and the answers of the plurality of learners, the degree of understanding and the correct answer rate of the learners of the test item It is characterized by including an analysis unit that analyzes the discriminating power of a problem from the item characteristics according to the above, and a transmission unit that transmits a problem that satisfies a predetermined condition according to the discriminating power of the analyzed item characteristics as a review question. do.

The program according to the present invention collects answers to test items by a plurality of learners, stores the collected answers of the plurality of learners in a storage unit, and based on the answers of the plurality of learners, the test items. The process of analyzing the discriminating power of a question from the item characteristics according to the learner's understanding level and the correct answer rate, and sending the question satisfying a predetermined condition according to the discriminating power in the analyzed item characteristics as a review question. It is characterized by having a computer execute it.

According to the present invention, efficient learning can be supported.

It is a schematic diagram which shows the configuration example of the learning support system. It is a block diagram which shows the configuration example of a server. It is explanatory drawing which shows an example of the record layout of a learner DB. It is explanatory drawing which shows an example of the record layout of the answer DB. It is explanatory drawing which shows an example of the record layout of a problem DB. It is explanatory drawing which shows an example of the IRT analysis result. It is explanatory drawing for demonstrating the extraction process of the review question according to the item characteristic. It is explanatory drawing for demonstrating transmission timing of a review question. It is explanatory drawing for demonstrating the transmission process of a review question. It is a flowchart which shows an example of the processing procedure which a server executes. It is a block diagram which shows the configuration example of the server which concerns on Embodiment 2. It is explanatory drawing which shows an example of the record layout of the transmission table which concerns on Embodiment 2. FIG. It is explanatory drawing for demonstrating transmission timing of the review question which concerns on Embodiment 2. It is a flowchart which shows an example of the processing procedure executed by the server which concerns on Embodiment 2. FIG. It is explanatory drawing which shows an example of the record layout of the transmission table which concerns on Embodiment 3. FIG. It is a flowchart which shows an example of the processing procedure executed by the server which concerns on Embodiment 3. FIG. It is explanatory drawing which shows an example of the record layout of the transmission table which concerns on Embodiment 4. FIG. It is explanatory drawing for demonstrating transmission processing of the review question which concerns on Embodiment 4. It is a flowchart which shows an example of the processing procedure executed by the server which concerns on Embodiment 4. FIG. It is a functional block diagram which shows the operation of the server of the above-mentioned form.

Hereinafter, the present invention will be described in detail with reference to the drawings showing the embodiments thereof.
(Embodiment 1)
FIG. 1 is a schematic diagram showing a configuration example of a learning support system. In this embodiment, a learning support system that distributes review questions for reviewing test items will be described as an example. The learning support system includes a learning support device 1 and terminal devices 2, 2, 2, .... Each device is communicated and connected via a network N such as the Internet.

The learning support device 1 is an information processing device that performs various information processing and also transmits / receives information, such as a server device and a personal computer. In the present embodiment, the learning support device 1 is assumed to be a server device, and is referred to as a server 1 in the following for convenience. The server 1 stores the results of the tests taken by the learners in a database, performs IRT analysis on the test results, and performs a process of evaluating the academic ability (understanding level) of each learner. The test is a test that is regularly conducted in, for example, an elementary school, a junior high school, etc., and is a test consisting of a plurality of test items (small questions). The server 1 collects the results of the tests conducted at each school, executes IRT analysis on the collected test results, and measures the academic ability of each learner. The server 1 distributes a later-described individual form showing the analysis result and a review question of the test according to the analysis result to each learner (see FIG. 9).

The terminal device 2 is a terminal device used by an instructor who guides a learner, and is, for example, a personal computer, a tablet terminal, or the like. In the present embodiment, the terminal device 2 is a personal computer. The instructor is, for example, a teacher at each school and scores the tests taken by the learner. The terminal device 2 transmits the answer data graded by the instructor to the server 1. The server 1 performs IRT analysis based on the answer data transmitted from the terminal device 2.

For example, the terminal device 2 may be a terminal operated by each learner, and the server 1 may provide a digital test so that each learner can take the test, as in e-learning.

FIG. 2 is a block diagram showing a configuration example of the server 1. The server 1 includes a control unit 11, a storage unit 12, a communication unit 13, and a large-capacity storage device 14.
The control unit 11 includes an arithmetic processing unit such as a CPU (Central Processing Unit) and an MPU (Micro-Processing Unit), and reads and executes a program P stored in the storage unit 12 to obtain various information related to the server 1. Performs processing, control processing, etc. Although the control unit 11 is shown as a single processor in FIG. 2, the control unit 11 may be a multiprocessor. The storage unit 12 includes memory elements such as a RAM (Random Access Memory) and a ROM (Read Only Memory), and stores a program P or data necessary for the control unit 11 to execute processing. Further, the storage unit 12 temporarily stores data and the like necessary for the control unit 11 to execute arithmetic processing. The communication unit 13 includes a processing circuit for performing processing related to communication, and transmits / receives information to / from terminal devices 2, 2, 2, ... Via the network N.

The large-capacity storage device 14 is a large-capacity storage device including, for example, a hard disk. The large-capacity storage device 14 stores the learner DB 141, the answer DB 142, and the question DB 143. The learner DB 141 stores data about a learner who takes a test item. The answer DB 142 stores data related to the answers to the test items by each learner. The question DB 143 stores the data of the review questions to be distributed to the learners according to the test results.

In the present embodiment, the storage unit 12 and the large-capacity storage device 14 may be configured as an integrated storage device. Further, the large-capacity storage device 14 may be composed of a plurality of storage devices. Further, the large-capacity storage device 14 may be an external storage device connected to the server 1.
Further, in the present embodiment, the server 1 is not limited to the above configuration, and may include, for example, an input unit that accepts operation input, a display unit that displays information related to the server 1, and the like.

FIG. 3 is an explanatory diagram showing an example of the record layout of the learner DB 141. The learner DB 141 includes a learner ID column, a name column, a school column, a grade column, a class column, and a number column. The learner ID column stores identification information for identifying each learner. The name column, school column, grade column, class column, and number column are associated with the learner ID, and the name, school name, grade, class, and attendance number of each learner are stored.

FIG. 4 is an explanatory diagram showing an example of the record layout of the answer DB 142. The answer DB 142 includes a test ID column, a subject column, a learner ID column, an answer date / time column, an answer column, a score column, a correct answer rate column, and a comprehension level column. The test ID column stores identification information for identifying each test performed periodically. The subject column stores the subject of each test in association with the test ID. The learner ID column stores the learner IDs of the learners who have taken each test in association with the test IDs. The answer date / time column stores the date when each learner took the test, that is, the answer date / time when the answer was given, in association with the test ID and the learner ID. The answer sequence stores the answers for each question given in the test and the scoring results of the answers in association with the test ID and the learner ID. In FIG. 4, the case of a correct answer is shown as "1", and the case of an incorrect answer is shown as "0". The score column and the correct answer rate column are associated with the test ID and the learner ID, respectively, and store the score and the correct answer rate related to the test result of each learner. The comprehension column stores the comprehension level of each learner analyzed by the IRT analysis process described later in association with the test ID and the learner ID.

FIG. 5 is an explanatory diagram showing an example of the record layout of the problem DB 143. The problem DB 143 includes a problem ID, a test ID, a test item string, and a data string. The question ID stores identification information for identifying the review question. The test ID column stores the test ID of the test corresponding to each review question in association with the question ID. The test item column stores information about each test item given in the test in association with the test ID. Specifically, the test item column stores the problem number and type of each test item. The data string stores the data of the review question corresponding to each test item in association with the question ID, the test ID, and the test item.

FIG. 6 is an explanatory diagram showing an example of the IRT analysis result. FIG. 6 illustrates a graph showing the results of analysis by IRT, a so-called Item Characteristic Curve (ICC). In the graph shown in FIG. 6, the horizontal axis shows the learner's comprehension level, and the vertical axis shows the correct answer rate of the test items.

The outline of the process executed by the server 1 will be described below. The server 1 collects the test answers by each learner from the terminal devices 2, 2, 2, ... Of each instructor. Specifically, the server 1 collects the answer data graded by each instructor. The server 1 stores the collected answer data in the answer DB 142. For example, when the number of accumulated answers reaches a predetermined number necessary for ensuring the reliability of the analysis result, the server 1 analyzes the collected answers by IRT.

IRT is a method for analyzing the characteristics of evaluation items or the characteristics of a subject according to the reaction of the subject to the evaluation item group. In the present embodiment, the evaluation items are each test item in the test, and the subject is a learner. The server 1 analyzes each learner's answer (reaction) to each test item, and evaluates the characteristics of the test item and the characteristics of the learner.

Specifically, the server 1 performs the analysis on the assumption that the analysis result applies to a certain mathematical model. For example, the server 1 applies a two-parameter logistic model (2PLM) to the answer data and applies it to the following equation (1) for analysis.

D is a constant, θi is the comprehension level of each learner, aj is the discriminating power of the test item, bj is the difficulty level of the test item, and Pj is the correct answer rate of the test item. Further, the reference numeral i indicates a code for identifying the learner, and the reference numeral j indicates a code for identifying the test item.
The server 1 specifies the discriminating power aj, the difficulty level bj, and the comprehension level θi that satisfy the equation (1) from the data group of the answers by the plurality of answerers. For example, the server 1 linearizes the discrete answer data using the PROX method, and specifies each parameter by using the maximum likelihood estimation method, the Bayesian estimation method, and the like. Since the algorithm is known, detailed description thereof will be omitted.

By IRT analysis, the server 1 derives the probability Pj that the learner with the comprehension degree θi correctly answers the j-th test item. FIG. 6 is a graph showing an example of the analysis result of the derived IRT analysis. Note that "Q1" is calculated with aj = 1 and bj = -1, "Q2" is calculated with aj = 0.5 and bj = -1, and "Q3" is calculated with aj = 1 and bj = 1. ..

According to the IRT, the item characteristic curve showing the item characteristic of the test item ideally draws an S-shaped curve as shown in FIG. Specifically, when the comprehension rate θi is low, the rate of increase in the correct answer rate Pj is small, but when the level of comprehension θi approaches the difficulty level bj, the rate of increase in the correct answer rate Pj increases, and when the level of comprehension θi becomes higher, the correct answer is given. The rate of increase in rate Pj decreases again. That is, when the comprehension degree θi is close to the difficulty level bj, the correct answer rate Pj rises.

The discriminating power aj represents the good discriminatingness of the test item as an evaluation scale. Specifically, the larger the discriminating power aj, the steeper the rise of the correct answer rate Pj when the comprehension degree θi is closer to the difficulty level bj. For example, as shown in FIG. 6, the correct answer rate Pj of "Q1" with aj = 1 has a difficulty level bj = with a degree of comprehension θi as compared with the correct answer rate Pj of "Q2" with aj = 0.5. The rate of increase is large when it is close to -1. As a result, the item characteristic curve of "Q1" draws a sharp S-shaped curve. Therefore, a learner with a high comprehension degree θi and a learner with a low comprehension degree θi can be well distinguished. Generally, the discriminating power aj takes a value of about 0.3 to 2.0.

The difficulty level bj represents the difficulty of the test item. The higher the difficulty level bj, the higher the numerical value of the comprehension degree θi at which the correct answer rate Pj rises. For example, as shown in FIG. 6, the correct answer rate Pj of “Q3” with bj = 1 rises around θi = 1, but the correct answer rate Pj of “Q1” with bj = -1 is θi = −. Stand up around 1. That is, the higher the difficulty level bj, the more the item characteristic curve is located on the right side of the figure, and the analysis result shows that the problem is more difficult. The difficulty level bj can be either positive or negative depending on the difficulty level of the test item.

As described above, the server 1 performs IRT analysis to obtain parameters aj and bj indicating not only the academic ability (understanding degree θi) of each learner but also the characteristics of each test item (small question) related to the conducted test. Identify. In the above, 2PLM is adopted as the mathematical model related to IRT, but the server 1 may adopt a one-parameter logistic model (1PLM, so-called rush model) focusing only on the difficulty level bj of the test item. Further, the server 1 may adopt a three-parameter logistic model (3PLM) in consideration of guesswork in addition to the discriminating power aj and the difficulty level bj.

FIG. 7 is an explanatory diagram for explaining the extraction process of the review question according to the item characteristics. In the present embodiment, the server 1 evaluates the validity of the question according to the item characteristics of each test item (small question), provides the learner with a review question only for the test item appropriate for review, and reviews. To do.
FIGS. 7A to 7D show four patterns of item characteristic curves. The item characteristic curve of FIG. 7A draws an S-shaped curve similar to the item characteristic curve shown in FIG. That is, the learner with a high correct answer rate Pj and the learner with a low correct answer rate Pj are well distinguished according to the comprehension degree θi, and the comprehension degree θi and the correct answer rate Pj are ideally correlated.

On the other hand, in the item characteristic curve of FIG. 7B, since the difficulty level bj is extremely low, the correct answer rate Pj is generally high regardless of whether the comprehension level θi is high or low. In other words, the test items do not function as an evaluation scale because many learners answered correctly because the test items were too simple.

Further, in the item characteristic curve of FIG. 7C, since the discriminating power aj is extremely small, the correct answer rate Pj is uniform regardless of whether the comprehension degree θi is high or low. That is, the distinctiveness of the test items is poor, and the difference in learner's ability cannot be properly discriminated.

Further, in the item characteristic curve of FIG. 7D, since the discriminating power aj takes a negative value, the higher the comprehension degree θi, the lower the correct answer rate Pj. In this case, for example, the test item is a so-called catching problem, and it is considered that the more capable learners are, the more incorrect the answer is. In the analysis results shown in FIG. 7D, the original purpose of the test of measuring the learner's academic ability is lost.

As mentioned above, there are cases where the problem is poor and the learner's academic ability cannot be measured appropriately. Even if the learner reviews such test items, there is a high possibility that they will not be able to properly support the improvement of academic ability. Therefore, the server 1 selects the test items to be reviewed by the learner according to the item characteristics of the test items identified by the IRT analysis. For example, the server 1 selects only test items whose discriminating power aj and difficulty level bj are in an appropriate numerical range as test items to be reviewed. For example, the server 1 selects test items having a discriminating power aj of 0.3 or more and 2.0 or less. Further, for example, the server 1 selects test items having a difficulty level bj of −3.0 or higher and 3.0 or lower.

The server 1 extracts the review questions of the selected test items from the question DB 143 and distributes them to each learner. For example, the server 1 delivers to each learner a review question of the test item that each learner made a mistake among the test items for which an appropriate analysis result was obtained.

FIG. 8 is an explanatory diagram for explaining the transmission timing of the review question. FIG. 8 shows a graph of fixation degree Qi showing the ratio of each learner remembering the test contents in chronological order. In the graph shown in FIG. 8, the horizontal axis represents the elapsed time t (week) from the answer date and time, and the vertical axis represents the fixation degree Qi. The server 1 transmits the review question extracted from the question DB 143 to each learner at the timing when the fixation degree Qi shown in FIG. 8 becomes equal to or less than a predetermined value.

The curve shown in FIG. 8 is a curve that approximates the forgetting curve of Ebbinghaus. The forgetting curve is a time-series curve representing forgetting of medium-term memory or long-term memory, and it is empirically known that the stored contents are attenuated in an exponential manner. According to the Ebbinghaus experiment, human memory decays exponentially, but by re-memory (review) when it is forgotten to some extent, the memory content becomes hard to forget.

In the present embodiment, the server 1 estimates the fixation degree Qi by reflecting the information of each learner in the function approximated by the above-mentioned forgetting curve. Specifically, the server 1 calculates the retention Qi for each learner by the following formula (2) according to each learner.

k represents a constant of 0 or more and less than 1, and t represents the elapsed time from the answer date and time to the present. αi is a coefficient according to the correctness of each learner's answer, and is, for example, the correct answer rate obtained by dividing the number of correct answers in the entire test by the number of test items (number of small questions).

Based on the equation (2), the server 1 estimates the retention Qi for each learner by multiplying the approximate equation k ^ t related to the forgetting curve by the coefficient αi according to the correctness of each learner's answer. .. For example, the server 1 uses a coefficient αi as the correct answer rate of the test by each learner. Further, the server 1 refers to the answer date and time stored in the answer DB 142, and calculates the elapsed time t for each learner. The server 1 calculates the retention degree Qi for each learner from the equation (2) based on the coefficient αi and the elapsed time t for each learner.

FIG. 8 shows the retention Qi of each of the learners A, B, and C. In the graph shown in FIG. 8, the calculation was performed with k = 0.5. Further, the calculation was performed with αi = 0.9 for the learner A, αi = 0.6 for the learner B, and αi = 0.7 for the learner C. Further, it is assumed that the learners A and B have taken the test at t = 0, and the learners C have taken the test at t = 2.

As shown in FIG. 8, the server 1 estimates that the fixation degree Qi is exponentially attenuated according to the elapsed time t. In this case, the server 1 estimates the fixation degree Qi according to each learner's grade (correctness of answer). For example, in FIG. 8, when the learners A and B are compared, the retention Qi of the learner A with good grades is higher than the retention Qi of the learner B with poor grades than the learner A. In this way, the server 1 calculates the fixation degree Qi according to the grade of each learner by determining the coefficient αi according to the correctness of the answer of each learner.

Further, the server 1 calculates the fixation degree Qi based on the answer date and time when each learner took the test. For example, when the learners A and C are compared in FIG. 8, the server 1 calculates the fixation degree Qi for the learner A according to the elapsed time t from t = 0. On the other hand, for the learner C, the server 1 calculates the fixation degree Qi according to the elapsed time t-2 from t = 2, which is two weeks behind the learner A.

For example, the server 1 calculates the fixation degree Qi for each learner every other week, and determines whether or not the fixation degree Qi is equal to or less than a predetermined value. The predetermined value is, for example, 0.4. The server 1 sends a review question at the timing when the retention Qi of each learner becomes 0.4 or less. For example, in the example of FIG. 8, the server 1 transmits the review question to the learner A at the timing of t = 3 when the fixation degree Qi is 0.4 or less. Similarly, the server 1 transmits the review questions to the learners B and C at the timings of t = 2 and 4, respectively.

As described above, the server 1 transmits the review question at the timing when the learner's retention Qi is lowered, that is, at the timing when the test content is almost forgotten. According to the Ebbinghaus experiment, by having the learner review the test content (memory content) at the timing when he / she is about to forget it, the memory can be efficiently fixed in a short review time.

In the above, the review question is simply transmitted at the timing when the fixing degree Qi becomes equal to or less than the predetermined value, but the present embodiment is not limited to this. For example, the server 1 may transmit the review question at the timing when the attenuation rate of the fixing degree Qi (the first derivative of Qi) becomes a predetermined value or more and the attenuation becomes gradual. That is, the server 1 only needs to be able to transmit the review question at the timing corresponding to the fixation degree Qi, and the transmission timing is not limited to the timing when the fixation degree Qi becomes a predetermined value or less.

FIG. 9 is an explanatory diagram for explaining the transmission process of the review question. The server 1 transmits the data shown in FIG. 9 to the learner at the timing determined above. Specifically, the server 1 sends an individual sheet showing the test result (upper part of FIG. 9) and a review question (“recommendation sheet” at the lower part of FIG. 9). The server 1 may send individual votes and review questions to the instructor's terminal device 2, or may directly distribute them to each learner. As shown in FIG. 9, the individual form is an instruction to tackle the review question (the part shown by the ellipse in FIG. 9) in addition to the comment automatically generated according to the scoring result of each learner's test and the grade. including.

For example, the server 1 provides the learner with three review questions per test. Specifically, as described above, the server 1 provides a review question corresponding to the test item (small question) that the learner made a mistake. For example, in the example shown in FIG. 9, the server 1 provides a review question of Q8 and Q10 that the learner made a mistake. Since Q8 and Q10 alone do not reach 3 questions, Server 1 also provides a review question of "B1" as a development question.

On the other hand, in the example shown in FIG. 9, since it is determined that the item characteristics of Q4 are not within an appropriate numerical range, the server 1 does not provide a review question regarding Q4. In this way, the server 1 appropriately selects the test items to be reviewed according to the item characteristics, and provides the review questions.

FIG. 10 is a flowchart showing an example of a processing procedure executed by the server 1. The processing contents executed by the server 1 will be described with reference to FIG.
The control unit 11 of the server 1 collects answers to test items by a plurality of learners from the terminal devices 2, 2, 2, ... (Step S11). For example, the control unit 11 collects answer data graded by each instructor. The control unit 11 stores the collected answers of each learner in the answer DB 142 in association with the answer date and time when each learner gave an answer (step S12).

The control unit 11 performs IRT analysis for analyzing the item characteristics of the test items and the degree of understanding of each learner based on the collected answers of the plurality of learners (step S13). Specifically, as described above, the control unit 11 analyzes the scoring result by applying it to a mathematical model such as 2PLM, and analyzes the item characteristics such as the discriminating power aj of the test item, the difficulty level bj, and the learner's comprehension degree θi. do.

The control unit 11 extracts the review questions to be provided to the learner from the question DB 143 according to the item characteristics of the analyzed test items (step S14). For example, the control unit 11 determines that the test item should be reviewed when the discriminating power aj is within a predetermined numerical range. Further, for example, when the difficulty level bj is within a predetermined numerical range, the control unit 11 determines that the test item should be reviewed. As a result, the control unit 11 identifies the test items on which the item characteristic curve draws an ideal S-shaped curve, and extracts the review questions only for the specified test items.

The control unit 11 estimates the current fixing degree of the test item for each learner according to the correctness of each learner's answer and the elapsed time from the answer date and time of each learner (step S15). Specifically, as described above, the control unit 11 calculates the fixation degree Qi for each learner by multiplying the approximate expression related to the forgetting curve of the Ebbinghaus by the coefficient αi according to the correctness of the answer.

The control unit 11 determines whether or not it is time to transmit the review question according to the current fixation degree estimated in step S15 (step S16). For example, the control unit 11 determines whether or not the fixing degree Qi is equal to or less than a predetermined value. If it is determined that it is not the timing to transmit (S16: NO), the control unit 11 returns the process to step S15. When it is determined that it is the timing to transmit (S16: YES), the control unit 11 transmits the review question extracted in step S14 (step S17), and ends a series of processes.

In the above, the server 1 is supposed to acquire the answer data scored by each instructor from the terminal device 2, but the present embodiment is not limited to this. For example, the server 1 may collate the answer collected from the terminal device 2 with the model answer and perform scoring. That is, the server 1 may collect unscored answer data.

Further, although the server 1 has decided to provide the review question only for the test item that the learner made a mistake in the above, the server 1 may also provide the review question for the test item that the learner answered correctly.

Further, in the above, it was decided to prepare a question for review in the question DB 143 separately from the question given in the test and deliver the question for review to the terminal device 2, but this embodiment is based on this. Not limited. For example, the server 1 may deliver the question given in the test as it is as a review question. That is, the server 1 determines whether or not the question is an ideal problem by IRT analysis, and if the problem that the learner makes a mistake is the ideal problem, the data of the problem has a reduced degree of fixation Qi. It is delivered to the terminal device 2 at the same timing. In this way, the test question and the review question may be the same.

From the above, according to the first embodiment, it is possible to select appropriate questions for review according to the item characteristics of the test items and have the learner review them, so that efficient learning is supported. Can be done.

Further, according to the first embodiment, a review question is prepared in the question DB 143 separately from the test question, and the review question is solved to support more efficient learning. Can be done.

Further, according to the first embodiment, by estimating the fixation degree Qi for each learner, it is possible to promote the review at an appropriate timing and to aim for more efficient fixation of the learning content.

(Embodiment 2)
In this embodiment, a mode for determining the timing of transmitting the review question according to the attributes of the learner will be described. The contents overlapping with the first embodiment are designated by the same reference numerals and the description thereof will be omitted.
FIG. 11 is a block diagram showing a configuration example of the server 1 according to the second embodiment. The server 1 according to the present embodiment stores the transmission table 144. The transmission table 144 is a table used when determining the timing at which the review question is transmitted.

FIG. 12 is an explanatory diagram showing an example of the record layout of the transmission table 144 according to the second embodiment. The transmission table 144 includes a class comprehension column and a coefficient column. The class comprehension column stores the comprehension degree θc in the entire class of the school to which the learner belongs. For example, the class comprehension degree θc is the comprehension level of all the learners (students) belonging to the class. The coefficient sequence stores the calculation method of the coefficient αi defined according to the class comprehension degree θc in association with the class comprehension degree θc. For example, the transmission table 144 defines that the coefficient αi becomes smaller as the class comprehension θc becomes lower for a class in which the class comprehension θc is a certain value or less (in the present embodiment, θc is 0 or less). That is, the transmission table 144 stipulates that the coefficient αi of the poor grade class is small.

FIG. 13 is an explanatory diagram for explaining the transmission timing of the review question according to the second embodiment. In the present embodiment, the server 1 determines the timing of transmitting the review question according to the learner's attributes, in addition to the correctness of each learner's answer and the elapsed time from the answer date and time.

The learner's attributes are, for example, information about a group such as a class, grade, or school to which the learner belongs. The server 1 refers to the learner DB 141 and determines the group to which each learner belongs. The server 1 weights the retention Qi according to which group each learner belongs to.

For example, the server 1 refers to the answer DB 142 and calculates the average value of the correct answer rate Pj in the entire class to which the learner belongs. The server 1 refers to the analyzed IRT analysis result (item characteristic curve) and specifies the class comprehension degree θc from the average value of the correct answer rate Pj. The server 1 estimates the current retention Qi of each learner based on the identified class comprehension θc. For example, the server 1 refers to the transmission table 144 and calculates the coefficient αi of the fixing degree Qi according to the class comprehension degree θc. Specifically, when the class comprehension degree θc is equal to or less than a certain value, the server 1 calculates the coefficient αi by multiplying the numerical value corresponding to the class comprehension degree θc by the correct answer rate of each learner. More specifically, the server 1 calculates so that the coefficient αi becomes smaller as the class comprehension degree θc is lower (see FIG. 12). As a result, the server 1 calculates so that the retention Qi is low for the learners in the class with low grades.

In FIG. 13, a graph comparing the retention Qi of the learner A belonging to the class having a high class comprehension θc and the retention Qi of the learner B belonging to the class having a lower class comprehension θc than the class of the learner A. Is shown. In the graph of FIG. 13, it is assumed that the grades of learners A and B are the same, and for learner A, αi = 1.0 (numerical value according to the class comprehension θc) × 0.7 (learner A). (Correct answer rate), and for learner B, αi = 0.6 × 0.7.

As shown in FIG. 13, the grades of the learners A and B are the same, but the retention Qi of the learner B belonging to the class having the class comprehension θc of a certain value or less is the retention Qi of the learner A. Is higher than. As a result, the timing of transmitting the review question to the learner B is earlier than the timing of transmitting the review question to the learner A. In the example of FIG. 13, the server 1 sends the review question to the learner B one week after the test is performed, and to the learner A two weeks later. In this way, the server 1 changes the timing of delivering the review question according to the attributes of each learner. This will prevent learning delays due to the learner's learning environment.

FIG. 14 is a flowchart showing an example of a processing procedure executed by the server 1 according to the second embodiment. After extracting the review questions according to the IRT analysis result (step S14), the control unit 11 of the server 1 executes the following processing.
The control unit 11 estimates the current degree of fixation of the test item for each learner according to the attributes of each learner (step S201). For example, the control unit 11 estimates the retention Qi according to the group such as the class, grade, and school to which each learner belongs. Specifically, the control unit 11 refers to the answer DB 142 and specifies the class comprehension degree θc in the entire class to which the learner belongs. The control unit 11 refers to the transmission table 144 and calculates the coefficient αi according to the specified class comprehension degree θc. Specifically, the control unit 11 calculates so that the coefficient αi becomes small for the learner belonging to the class having a low class comprehension degree θc. As a result, the control unit 11 estimates the retention Qi of the learner in the low grade class to be low. The control unit 11 shifts the process to step S16.

In the above, the fixation degree Qi is estimated according to the class comprehension degree θc, but the present embodiment is not limited to this. For example, the server 1 may estimate that the degree of fixation Qi is lowered by reducing the coefficient αi as the learner's grade is higher. As a result, the timing of reviewing the upper grades learners (students) is advanced, and the timing of reviewing the lower grades learners is delayed. As described above, the server 1 only needs to be able to estimate the fixation degree Qi according to the attributes of the learner, and the estimation method of the fixation degree Qi is not particularly limited.

Further, in the above, the fixation Qi is weighted only for the class in which the class comprehension θc is below a certain value, that is, the class with poor grades, but the fixation Qi is also weighted for the class in which the class comprehension θc is above a certain value. You may go.

From the above, according to the second embodiment, the review timing can be adjusted according to the learning environment and the situation by estimating the fixation degree Qi according to the attributes of the learner.

(Embodiment 3)
In the second embodiment, the timing of transmitting the review question is determined according to the attributes of the learner. In this embodiment, a mode for determining the timing of transmitting the review question according to the attributes of the test items will be described.
FIG. 15 is an explanatory diagram showing an example of the record layout of the transmission table 144 according to the third embodiment. The transmission table 144 according to the present embodiment stores the subject comprehension degree θs instead of the class comprehension degree θc. The subject comprehension degree θs is the learner's comprehension degree θs for each subject analyzed from the test of each subject. The transmission table 144 stores a calculation method of the coefficient αi according to the subject comprehension degree θs in association with the subject comprehension degree θs. For example, the transmission table 144 defines that when the subject comprehension degree θs is a certain value or less (in the present embodiment, θs is 0 or less), the coefficient αi becomes smaller as the subject comprehension degree θs is lower. That is, the transmission table 144 stipulates that the coefficient αi is small for the fixing degree Qi of the test of the weak subject.

The server 1 calculates the retention Qi according to the attributes of the test taken by the learner. Specifically, the server 1 refers to the transmission table 144 and calculates the fixation degree Qi for each subject. For example, the server 1 calculates the subjects that it is not good at (subjects with θs of 0 or less) so that the coefficient αi becomes small, and underestimates the fixing degree Qi. As a result, the server 1 accelerates the transmission timing of the review questions related to the subjects that it is not good at. By transmitting the review questions of the weak subjects at an early stage, the server 1 aims to prevent the learning trips in the weak subjects. Since it is almost the same as the second embodiment except that the subject comprehension degree θs is adopted instead of the class comprehension degree θc, detailed illustration and description will be omitted in this embodiment.

FIG. 16 is a flowchart showing an example of a processing procedure executed by the server 1 according to the third embodiment. After extracting the review questions according to the IRT analysis result (step S14), the control unit 11 of the server 1 executes the following processing.
The control unit 11 estimates the current fixing degree of the test item for each learner according to the attribute of the test item (step S301). For example, the control unit 11 estimates the retention Qi according to the subject of the test taken by each learner. Specifically, the control unit 11 refers to the answer DB 142 and specifies the learner's degree of understanding θs for each subject for each subject. The control unit 11 refers to the transmission table 144 and calculates the coefficient αi according to the subject comprehension degree θs. Specifically, when the subject comprehension degree θs is equal to or less than a certain value, the control unit 11 calculates so that the lower the subject comprehension degree θs, the smaller the coefficient αi. As a result, the control unit 11 estimates the degree of fixation Qi of the subject that it is not good at low. The control unit 11 shifts the process to step S16.

In the above, the retention Qi is estimated for each subject, but the present embodiment is not limited to this. For example, the server 1 may estimate the fixation degree Qi according to the type of test item (for example, problem type). Specifically, the server 1 performs IRT analysis for each type of problem, such as a problem asking "knowledge", a problem asking "skill", and a problem asking "idea", and analyzes the learner's comprehension degree θi for each type. do. The server 1 calculates the coefficient αi based on the comprehension degree θi according to the type of the problem. In this way, the server 1 only needs to be able to estimate the fixation degree Qi according to the attributes of the test items, and the attributes of the test are not limited to the subjects.

From the above, according to the third embodiment, by estimating the fixation degree Qi according to the attributes of the test items, it is possible to encourage the review at the timing according to the learner's ability.

(Embodiment 4)
In this embodiment, a mode for determining the transmission timing of the review question according to the analysis result of the IRT will be described.
FIG. 17 is an explanatory diagram showing an example of the record layout of the transmission table 144 according to the fourth embodiment. The transmission table 144 according to the present embodiment defines the coefficient αi according to the item characteristics of the test items. For example, the transmission table 144 defines the coefficient αi according to the discriminating power aj of the test item. Specifically, the transmission table 144 stipulates that the coefficient αi becomes smaller as the discriminating power aj approaches the ideal numerical range of 0.3 to 2.0. As a result, the server 1 calculates the fixation Qi lower as the discriminating power aj is closer to the ideal numerical range, and sends the review question to the learner at an earlier timing.

FIG. 18 is an explanatory diagram for explaining the transmission process of the review question according to the fourth embodiment. In the first embodiment, the server 1 extracts the review questions only for the test items whose item characteristics are in the ideal numerical range and sends them to the learner. On the other hand, in the present embodiment, the server 1 weights according to the item characteristics of the test items, and preferentially transmits the review questions from the test items having good item characteristics.

Specifically, the server 1 performs IRT analysis on the answers of each test item given in the test, and identifies item characteristics such as discriminating power aj and difficulty level bj. As a result, the server 1 obtains the item characteristic curve of each test item as shown in FIG. Here, as shown in FIG. 18, the discriminating power aj is in the ideal numerical range, and the test item in which the curve draws an S-shaped curve (upper left in FIG. 18) and the discriminating power aj are from the ideal numerical range. It is assumed that there are test items that are slightly off and have a gentle curve (middle row on the left in FIG. 18) and test items that have extremely small discriminating power aj and have no discriminating power (lower row on the left in FIG. 18). In this case, the server 1 determines the transmission timing of the review question and the presence / absence of transmission according to the discriminating power aj of each test item.

For example, the server 1 refers to the transmission table 144, weights the coefficient αi for the ideal test item shown in the upper part of FIG. 18, and estimates the fixation Qi low. As a result, the server 1 sends the review question at an early timing. On the other hand, the server 1 calculates the test items shown in the middle part of FIG. 18 so that the coefficient αi is larger than the test items in the upper part of FIG. 18 because the discriminating power aj is out of the ideal numerical range. As a result, the server 1 sends the review question to the learner later than the test item in the upper part of FIG. Further, the server 1 does not adopt the test item in the lower part of FIG. 18 as a test item to be reviewed because it does not have the discriminating power aj, and does not transmit the review question. In this way, the server 1 determines the transmission timing of the review question and the presence / absence of transmission according to the degree of whether or not the item characteristic of the test item is an ideal numerical value. As a result, review is given priority to test items that are appropriate for review, and more efficient learning can be supported.

FIG. 19 is a flowchart showing an example of a processing procedure executed by the server 1 according to the fourth embodiment. After extracting the review questions according to the IRT analysis result (step S14), the control unit 11 of the server 1 executes the following processing.
The control unit 11 estimates the current fixing degree of the test item according to the item characteristics of the test item analyzed in step S13 (step S401). For example, the control unit 11 estimates the fixing degree Qi according to the discriminating power aj of the test item. Specifically, the control unit 11 refers to the transmission table 144 and calculates so that the coefficient αi becomes smaller as the discriminating power aj is closer to the ideal numerical range. As a result, the control unit 11 calculates the fixing degree Qi of the test item having good item characteristics to be low. The control unit 11 determines whether or not it is the timing to transmit the review question according to the fixing degree estimated in step S401 (step S402). For example, the control unit 11 determines whether or not the fixing degree Qi is equal to or less than a predetermined value. If it is determined that it is not the timing to transmit (S402: NO), the control unit 11 returns the process to step S401. When it is determined that it is the timing to transmit (S402: YES), the control unit 11 transmits a review question of the test item whose fixing degree is equal to or less than a predetermined value (step S403).

The control unit 11 determines whether or not transmission has been completed for all the review questions extracted in step S14 (step S404). If it is determined that the transmission has not been completed (S404: NO), the control unit 11 returns the process to step S401. As a result, the control unit 11 recalculates the current fixation degree Qi for the test items for which the review question has not been transmitted, and transmits the review question at any time when the fixation degree Qi becomes equal to or less than a predetermined value. When it is determined that the transmission of all the review questions has been completed (S404: YES), the control unit 11 ends a series of processes.

From the above, according to the fourth embodiment, the test items to be reviewed can be ranked by estimating the fixing degree Qi according to the item characteristics of the test items.

(Embodiment 5)
FIG. 20 is a functional block diagram showing the operation of the server 1 in the above-described form. When the control unit 11 executes the program P, the server 1 operates as follows. The collecting unit 201 collects the answers to the test items by a plurality of learners. The storage unit 202 stores the collected answers of the plurality of learners. The analysis unit 203 analyzes the item characteristics of the test item based on the answers of the plurality of learners. The transmission unit 204 transmits a review question according to the analyzed item characteristics.

Since the fifth embodiment is as described above and the other parts are the same as those of the first to fourth embodiments, the corresponding parts are designated by the same reference numerals and detailed description thereof will be omitted.

The embodiments disclosed this time should be considered to be exemplary in all respects and not restrictive. The scope of the present invention is indicated by the scope of claims, not the above-mentioned meaning, and is intended to include all modifications within the meaning and scope equivalent to the scope of claims.

1 server (learning support device)
11 Control unit 12 Storage unit P program 13 Communication unit 14 Large-capacity storage device 141 Learner DB
142 Answer DB
143 Problem DB
144 Transmission table 2 Terminal device

Claims (8)

A collection department that collects answers to test items by multiple learners,
A storage unit that stores the collected answers of the plurality of learners,
Based on the answers of the plurality of learners, an analysis unit that analyzes the discriminating power of the problem from the item characteristics according to the learner's understanding level and the correct answer rate of the test item, and the analysis unit.
A learning support device including a transmission unit that transmits a question satisfying a predetermined condition according to the discriminating power in the analyzed item characteristics as a review question. A second storage unit that stores the review questions corresponding to the test items,
It is equipped with an extraction unit that extracts the review questions according to the item characteristics analyzed by the analysis unit.
The learning support device according to claim 1, wherein the transmission unit transmits the review question extracted by the extraction unit. The storage unit stores the answers of the plurality of learners in association with the answer date and time when each of the learners gave an answer.
Each learner is provided with an estimation unit that estimates the current degree of fixation of the test item based on the correctness of each learner's answer and the elapsed time from the answer date and time.
The learning support device according to claim 1 or 2, wherein the transmission unit transmits the review question at a timing corresponding to the degree of fixation. The learning support device according to claim 3, wherein the estimation unit estimates the degree of fixation according to the attributes of the learner. The learning support device according to claim 3 or 4, wherein the estimation unit estimates the degree of fixation according to the attributes of the test item. The learning support device according to any one of claims 3 to 5, wherein the estimation unit estimates the degree of fixation according to the item characteristics analyzed by the analysis unit. A learning support system having a learning support device and a terminal device that communicates with the learning support device.
The learning support device is
A collection unit that collects answers to test items by multiple learners from the terminal device,
A storage unit that stores the collected answers of the plurality of learners,
Based on the answers of the plurality of learners, an analysis unit that analyzes the discriminating power of the problem from the item characteristics according to the learner's understanding level and the correct answer rate of the test item, and the analysis unit.
A learning support system including a transmitter that transmits a question satisfying a predetermined condition according to the discriminating power in the analyzed item characteristics as a review question. Collecting answers to test items by multiple learners,
The collected answers of the plurality of learners are stored in the storage unit, and the answers are stored.
Based on the answers of the plurality of learners, the discriminating power of the problem is analyzed from the item characteristics according to the learner's understanding level and the correct answer rate of the test item.
A program characterized by having a computer execute a process of transmitting a question satisfying a predetermined condition according to the discriminating power in the analyzed item characteristics as a review question.

Images