KR102561666B1 - Efficiently study method of problem-bank presenting problems style using Artificial intelligence algorithm - Google Patents

Abstract

The present invention relates to an efficient learning method using a question bank type using an artificial intelligence algorithm. The present invention comprises the steps of: constructing a question bank DB; producing a mock test; and providing the mock test to a test taker's terminal.

Description

Efficient study method of problem-bank presenting problems style using Artificial intelligence algorithm}

The present invention relates to an efficient learning method using an artificial intelligence algorithm, and more specifically, to a question bank using an artificial intelligence algorithm that can improve skills and grades by receiving efficient learning suitable for oneself. It is about an effective learning method of formula presentation method.

Recently, as computer, electronic, and communication technologies have developed rapidly, online evaluation tests using computer and communication technologies have been spotlighted as a new evaluation method.

These online evaluation tests are gradually expanding their utilization range because they eliminate space limitations and make it easy to process results such as scoring.

Such an online evaluation test may be presented in the form of a question bank. The question bank collects many problems in one place, builds a database, and presents questions from the database, so it is possible to create questions automatically. In addition, the method of setting questions using a question bank is widely used because it is possible to promptly set questions, easily adjust the level of difficulty, and efficiently set evaluation tests.

The purpose of the evaluation test is to confirm the learning completion level of the learner, and the best way to confirm the learning completion level is through the evaluation of narrative or subjective questions. However, this method has a disadvantage in that it is not objective because it takes a lot of time and money as the number of learners increases, and the evaluation of descriptive or subjective questions may vary depending on the rater.

Therefore, in most online evaluation tests, the method of confirming the learning completion level of learners through multiple-choice questions has become common as an alternative to the evaluation of descriptive or subjective questions.

In the case of multiple-choice questions, it is possible to process grades for many learners in a short time through the OMR card, and there is an advantage in that the evaluation of the multiple-choice questions does not vary depending on the scorer.

On the other hand, since learning does not necessarily follow blindly hard work, that is, it is well known that how hard one works, not blindly hard work, is important.

Moreover, in the case of studying to solve a multiple-choice test, it is important to study by covering the areas in which you are strong and the areas in which you are weak. There were fatal disadvantages in that the learning efficiency was very low for students or examinees who had to study many subjects in a limited time because they were sometimes omitted or neglected.

In order to overcome these problems, not only learners' independent learning but also appropriate learning methods are required. The present invention also relates to a method for providing a learning method using online.

Therefore, in the question bank-style online learning method, classifying each learning category, reducing the amount of learning for each individual with a low percentage of errors due to the large amount of learning, and intensively learning the categories with a large amount of errors due to the insufficient amount of learning. The development of a system or learning method that can increase efficiency was very urgent.

Korean Patent Publication No. 2021-0051256 Korean Patent Publication No. 2020-0048594 Korean Patent Publication No. 2019-0125055

Therefore, the present invention is a method of classifying each learning category in a question bank-style online learning method, reducing the amount of learning for each individual with a low error rate due to the rich amount of learning, and intensively learning the category that is often wrong due to the insufficient amount of learning. It is about an efficient learning method using an artificial intelligence algorithm that can increase the efficiency and skill of learning.

In order to achieve this object, the present invention provides a service providing server that provides an online evaluation service to a test taker's terminal constructing a question bank DB including a plurality of question sets, Selecting at least one problem set, extracting problems from the problem set selected in the above step for each part and constructing multiple-choice options, constructing mock test questions through the service providing server, and ordering multiple-choice questions In the efficient learning method of the question bank question method using an artificial intelligence algorithm, including the step of producing a mock test by mixing and the step of providing the mock test to a test taker terminal,

The problem providing step of providing online a problem for the candidate accessing the service providing server to solve, the information receiving step of receiving information including the learner's marking content for the problem, and the information receiving step A marking item analysis step of analyzing the items marked by the learner through the information data conversion step of turning the information into data and the data analysis stored by the learning information conversion step; And after the step of providing analysis information on the marked item analyzed by the step of analyzing the marking item to the test taker and the step of learning based on the vulnerability through the step of providing the item analysis information and the step of providing the item analysis information. , Re-taking step of reconnecting to the service providing server and taking the test through the test taker terminal, and analyzing the re-taking result in the re-taking step and predicting pass or fail.

In addition, the analysis of the marking item analysis step is characterized in that if 30% or less of the entire test taker is correct, it is judged as A grade difficulty, 30 to 60% or less is correct, B grade difficulty, and 60% or more is determined as C grade difficulty. .

In addition, the marking question analysis step is characterized in that, through the detailed items of the candidate, if all parts of the candidate are wrong, they are classified as A1, parts that are partially correct and partly wrong are classified as B1, and parts that are all correct are classified as C1.

In addition, the marking item analysis step is characterized in that the parts classified as the A1, the B1, and the C1 are provided to the test taker so that the test taker can focus on learning.

In addition, if it is expected to obtain a passing score or higher in the result estimation step, it is characterized in that the mock test is retaken by selection.

Therefore, the present invention can use artificial intelligence technology using big data to intensively learn the weak parts of each candidate, so that the efficiency of learning can be improved, good grades can be expected in related qualification tests, and effective test-takers can improve their grades. There is an excellent effect that can promote.

In addition, in the question bank-type online learning method, the efficiency of online learning is a method of classifying each field of learning, reducing the amount of learning for each individual with a low percentage of errors due to the large amount of learning, and intensively learning the category with a large amount of errors due to the small amount of learning. There is also an effect that can increase the .

1 is a diagram showing the configuration of a system for realizing a question bank-type online evaluation test service.
2 is a flow chart of a method of providing an online evaluation service using a question bank method for providing an online evaluation test by using a question bank method.
Figure 3 is a flow chart of an efficient learning method of the question bank question method using an artificial intelligence algorithm according to the present invention.
4 is a diagram showing the configuration of an information analysis system;
5 is a flowchart showing a method for predicting an actual test pass rate;

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to components of each drawing, it should be noted that the same components have the same numerals as much as possible even if they are displayed on different drawings.

In addition, in the following description of the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description will be omitted.

In addition, since the terms used in this application are only used to describe specific embodiments, it is not intended to limit the present invention, and it is clear in advance that a singular expression also means a plurality of expressions unless the context clearly indicates otherwise. want to leave

Prior to describing the present invention, a computer appearing in this specification means hardware (or computing device) capable of executing a computer program, and the hardware (or computing device) includes a PC, a laptop computer, a tablet PC, and an Internet mobile device. (Internet Mobile Device (MID)), smart phone, processor, or server.

A computer program executed by hardware (or a computing device) may be implemented in software such as an application (or application program) or a mobile app. The computer program may automatically analyze the learner's multiple-choice questions and/or an answer (or correct answer) for each of the multiple-choice questions using an artificial intelligence (AI) algorithm.

1 is a diagram showing the configuration of a system for implementing a question bank-type online evaluation test service, and FIG. 2 is a flow chart of a question bank-type online evaluation service providing method that provides a question bank-type online evaluation test, Figure 3 is a flow chart of an efficient learning method using the question bank question method using an artificial intelligence algorithm according to the present invention, Figure 4 is a diagram showing the configuration of an information analysis system, Figure 5 is a flow chart showing a method for predicting the actual test pass rate .

Hereinafter, with reference to drawings, each component of a system for implementing a question bank-type evaluation test service will be described in detail.

Referring to FIG. 1 , the service providing server 10 may be a server that provides an online evaluation service to the candidate terminal 20 . The service providing server 10 builds a question bank DB 11 containing a plurality of question sets, randomly selects question sets to be set for individual test takers through the question bank DB 11, and produces an online evaluation test. It can be provided to the test taker terminal 20 through a network.

The service providing server 10 allows access of the test taker terminal 20 through a network including the Internet, an intranet, a wired/wireless communication network, a mobile communication network, and the like, and transmits and receives various signals and data.

In particular, the online evaluation test is provided to the test taker terminal 20 by providing different problem sets, multiple choice options, and item order for each test taker, so that test takers can take the online test fairly without memorizing questions or cheating. It allows you to objectively measure your own skills. Here, the test taker means an examinee who prepares for the test with a related test book in order to take the national qualification test.

The candidate terminal 20 may be a terminal of a candidate who wishes to take an online evaluation test using the online evaluation service providing method provided by the service providing server 10 . The test taker terminal 20 may be various communication terminals such as computers, laptop computers, PDAs, MP4 players, and mobile communication terminals, and may also be various smart devices such as smart phones, smart notes, tablet PCs, and smart TVs.

However, the test taker terminal 20 is not limited to the same terminal types as those listed above, and if it is possible to take an online evaluation test by accessing the service providing server 10 through a network, regardless of the specific terminal type, this Any role of the candidate terminal 20 of the embodiment can be played.

On the other hand, a smart device refers to a product whose functions are not limited and whose functions can be changed or expanded through application programs.

If the candidate terminal 20 is a smart device on which an application program capable of taking an online evaluation test is installed, the candidate can more easily take the online evaluation test and check the result.

The service providing server 10 connected to the test taker terminal 20 is also connected to the big data management unit 30 to store data related to the test taker's test taker history and test scores.

Hereinafter, the big data management unit 30 will be described.

Referring to FIG. 1 again, the big data management unit 30 stores the test taker's test taker's history and test scores, and data on the learning map of the test taker's weak part. It may include a unit 32, a difficulty analysis unit 33, a test result storage unit 34, and a personal learning unit 35, and is not limited to the above components, and all related to the test taker's test taker history and score results. may contain data.

The learning date unit 31 calculates the interval between the date the candidate took the mock test for the first time and took the next mock test, and determines whether the score has increased by studying for several days.

The grade rise judgment unit 32 judges how much the grade has increased according to the difficulty of the mock test in conjunction with the difficulty analyzer 33 to be described later. For example, if the score is almost similar to the previous mock test score but the difficulty is very high In fact, it is judged that the grade has risen, and even if the score has risen to a certain extent, if the difficulty level is very low, it is judged that the grade has not actually risen. Therefore, it would be desirable to generate numerical values or graphs of the entire ranking and transmit the results to the candidate. In other words, it becomes a data to judge the learning achievement of test takers.

The difficulty analysis unit 33 analyzes the percentage of correct answers of test takers who took the mock test, analyzes the overall difficulty level, and quantifies it. In addition, it is possible to display the current position of the candidate by comparing with the average score of all applicants using a polygonal line graph or the like.

The test result storage unit 34 can indicate the total number of correct answers for each test taker, whether errors (less than 40 points) have occurred, the overall average score, the average score of subjects, the probability of passing, the result of retaking the mock test, etc. will be.

The personal learning unit 35 analyzes the weak parts of each test taker who took the mock test, separately extracts the problem of the weak part, presents it to the candidate, and learns through the presentation of the problem of the weak part and its solution as a whole. is to do

Hereinafter, the operation of the system for realizing the question pool type evaluation test service will be described.

First, questions are extracted from the question bank DB 11 stored in the form of a question pool, and the service providing server 10 edits them in the form of a mock test to solve the mock test through the test taker terminal 20. .

Then, in the big data management unit 30, the learning date unit 31 counts the test taker's test take date and the number of test takers. The grade rise determination unit 32 can know the grade increase, and in this case, it is possible to determine the current position through the total ranking of the test takers. The difficulty level of the mock test is checked through the difficulty analyzer 33 and linked with the grade increase determination unit 32 to determine how much the mock test score has increased.

The test result storage unit 34 is also linked with the difficulty analysis unit 33 to calculate the accumulation of mock test scores, the average score, the average rank difference, and the probability of obtaining a passing score of 60 or more so that the test taker can provide the information to the test taker. and is stored.

In the personal learning unit 35, based on the scores for each part calculated through the test result storage unit 34, the weak part of each test taker who took the mock test is searched for, and that part is separately presented to the corresponding candidate, and the correct answer is given. Through the explanation of the test taker and the presentation of the solution method, it is to make up for the weakness by further learning the lacking part of the test taker.

Hereinafter, a method for providing an online evaluation service using a question bank method will be described with reference to drawings.

2 is a method of providing online evaluation using a question bank DB (11) method for providing mock exams, in which a question bank DB (11) including a plurality of question sets is constructed (S10), the service providing server (10) Selecting at least one problem set to be set through the question bank DB 11 (S20), extracting questions from the problem set selected in S20 for each part and constructing multiple choice options (S30), providing service Forming mock test questions through the server 10 (S40), creating a mock test by mixing the order of multiple choice questions (S60), and providing the mock test to the test taker terminal 20 (S60). can be implemented.

In the step S10, the service providing server 10 may build a question bank DB 11 including a plurality of question sets composed of questions, correct answers, and a plurality of incorrect answers. In this case, the problem set may include more incorrect answers than a predetermined number of incorrect answers constituting the multiple-choice option. That is, in the case of multiple-choice questions, incorrect answers in addition to the questions and correct answers can constitute a problem set. There may be.

Furthermore, in order to provide different options for each test taker, the problem set may be configured to include more wrong answers than a predetermined number. For example, a problem set may be configured with 1 correct answer, 10 incorrect answers, and the like, and stored in the question bank DB 11.

Meanwhile, among the questions stored in the question bank DB 11 constructed in step S10, there may be questions including fingerprints, and at least one question set may be assigned to one fingerprint.

Therefore, in step S10, a plurality of questions and question sets may be included, and a question bank DB 11 may be constructed by storing a plurality of question sets assigned to one fingerprint together. At this time, it is also possible to configure the fingerprint with text or images.

In step S20, the service providing server 10 may select at least one or more question sets to be set for individual test takers through the question bank DB 11. That is, problem sets to be set for each candidate may be different from each other, and the number of problem sets for producing an online evaluation test may be selected.

In this case, since the number of items may be determined according to the online evaluation test, in step S20, a set of questions corresponding to the determined number of items may be selected. In addition, in step S20, the online evaluation test of the same difficulty level may be provided to all test takers by selecting a set of problems according to a predetermined difficulty level.

In addition, in order to prevent cheating, since the questions that can be presented in one passage may vary even if the passage is the same, it is possible to select and select a set of questions so that the same passage is provided to all test takers while the detailed questions are different from each other. can

In step S30, the service providing server 10 may randomly extract questions by unit or part from the problem set selected in step S20 to configure multiple choice options separately. By configuring the extracted multiple-choice options, it is possible to provide different multiple-choice options for each test taker even for the same problem.

In the step S30, in the case of the 1st, 2nd, 3rd, and 4th choices of the 4-choice multiple-choice problem, multiple-choice options may be configured by making the order of the choices different from each other.

In this way, memorization of questions or cheating can be prevented by constructing multiple choice options by confusingly changing the order of viewing multiple choice questions consisting of 1, 2, 3, and 4.

In step S40, the service providing server 10 may compose mock test questions with multiple choice options configured differently from the questions included in the selected question set in the manner of step S30.

As described above, the arrangement of the multiple-choice questions constructed in step S40 may be different for each candidate, and even if the question is the same, the multiple-choice options may be different from each other.

In step S50, the service providing server 10 may create a mock test, which is an online evaluation test, by mixing the order of multiple-choice questions of the mock test constructed in step S40. Multiple-choice questions are formed according to the performance of steps S30 to S50 through the set of questions selected to be presented to individual candidates in step S20.

In addition, in step S50, by mixing the order of at least one or more multiple-choice questions configured in step S40 to create an online evaluation test, it is possible to solve the corresponding questions in a different order for each candidate even if it is the same item. In this way, cheating using regular signals or communication devices can be prevented by mixing the order of multiple-choice questions.

Furthermore, the S50 is not limited to the above-described multiple-choice questions, but is also applicable to subjective questions.

In step S60, the service providing server 10 may provide the produced mock test to the test taker terminal 20 of each test taker.

That is, by providing the mock test, which is the online evaluation test produced in step S50, to the test taker terminal 20 of each test taker, the test taker can take the online mock test evaluation. In the step S60, the prepared mock test may be transmitted to the test taker terminal 20 at once and provided, but the online evaluation test may be provided by sequentially transmitting a predetermined number of multiple-choice questions.

Through the above steps, the test taker solves the mock test provided through the test taker terminal 20 . Hereinafter, this will be described with reference to the drawings.

Referring to FIG. 3 , problems to be solved by the examinee who accesses the question bank providing server 10 are provided online (step 1).

The problem content of the question bank provided by the question bank providing server 10 is a learning question to be solved by the learner, and includes other learning contents such as problem solving.

As for the learning items, a plurality of learning items are sequentially provided according to a predetermined number and level of difficulty per session. For example, 10 questions may be provided collectively so that the learner scrolls through the solution, or a small number of questions such as one or two may be provided so that they can be checked by going over the screen. Therefore, it will be provided with a means for the learner to check the correct answer for solving the question.

In addition, a means for selecting the next question is also presented so that the learner can solve another question first by skipping the question presented to solve it later. The next page may be selected by a touch input method.

In addition, it may be provided that all questions of the corresponding round can be collectively scored, or that scoring can be performed for each question provided on one page.

Marking information for receiving information including the learner's marking content for the problem is received (step 2).

While solving the mock test provided by the mock test question providing step, which is the first step, if there is a marking for the question, the service providing server 10 receives the test taker's answer sheet marking. Since the learning information on the learning questions of the mock test may include the candidate's response to the question and the response time from the provision of the question to the answer, the time taken to solve each question can also be measured.

The marking information received by the marking information receiving step, which is the second step, is converted into data (third step).

In this way, by the learning information receiving step (step 2), the marking results for the questions of the test taker's mock test are transmitted to the big data unit 30 so that the service providing server 10 converts them into data.

In principle, the number of test takers taking the online mock test is infinite due to the nature of the online service, so the sample group is very large. In the third step, the marking information data generation step, the marking result information by numerous test takers is converted into big data, and analyzed to be described later. Through the analysis step of the marked item, which is the fourth step to be performed, specific analysis information on the marked item is provided in the fifth step, which is the marking item analysis and provision step.

In the step of providing marking question analysis information, of course, information on general learning results such as the total correct answer rate per round, total solving time, difficulty level, and comparison of average test taker scores, as well as the marking item analysis information according to the fourth step, are included.

In the 4th step, the marking item analysis step, the test taker's customized learning based on the marking information of the test taker received through the marking information receiving step of the 2nd step and the information that is converted into big data and held by the marking information data conversion step which is the 3rd step. After analyzing the questions, feedback information that can be helpful for the candidate's learning direction is analyzed to present problems.

Since the analysis of the marking questions is first to measure the difficulty of the mock test problem, for example, a correct answer rate of less than 30% is A grade difficulty, a correct answer rate of less than 30 to 60% is B grade difficulty, and a correct answer rate of 60% or more is C grade It is judged by difficulty.

In the marking item analysis step (fourth step) of analyzing the items marked by the candidate through the data analysis stored by the marking information dataization step corresponding to the third step, a specific candidate in a detailed item, such as the Korean Industrial Manpower Assume that this is a mock test for the ergonomic engineer qualification test administered by the Corporation (Q-net).

If the items of the above subjects are divided into 4 categories, it is assumed that they are divided into ergonomics introduction subjects, occupational physiology subjects, industrial psychology subjects, and musculoskeletal disorder prevention and management subjects.

Hereinafter, an example of an introductory ergonomics course among the above courses will be given.

If you get all the mock test questions wrong in the human information processing part, which is the detailed item of the Human Engineering Introduction subject, if you assume that you got some of the mock test questions right and some of them wrong in the very weak part and human machine system part, you can get the weak part, anthropometric and application part If you assume that you have answered all the mock test questions correctly, it is considered a strong part.

In this way, in the learning item analysis information providing step (fifth step), which provides analysis information on the learning item analyzed by the marking item analysis step, which is the fourth step, not only the subject analysis analyzed through the fourth step, but also the Assuming that the human information processing part, which is a detailed item, is supplemented first, and the partially correct and partially incorrect human machine system part is supplemented second, the feedback of these contents is transmitted to the applicable candidate.

In the fourth step, which is the marking question analysis step, through the detailed items of the candidate (human information processing, human machine system, etc.), if the candidate is all wrong, A1, part correct and part wrong, B1, all correct parts to be classified as C1. Therefore, it is possible to comment that it is necessary to study A1, part more intensively.

That is, the parts classified as A1, B1, and C1 are provided to the applicant, and since they are weak areas, they can be intensively studied and supplemented.

Through the fifth step, which is the marking analysis information providing step, learning is performed centering on vulnerabilities (sixth step).

In the above step, considering the difficulty level of the person's weak part, a video of problem solving related to problem solving through a question bank is provided.

In addition, in general, when studying such as taking an exam, the applicant is often short of subjects and continues to make wrong parts in particular, and wastes time having to study again for the perfect part with a very high correct answer rate. Test takers (students) overlooked these points and it was easy to learn, so it was difficult for test takers to improve their skills.

Therefore, by referring to the feedback presented through the above-mentioned 5th step, the test taker is allowed to focus on learning his or her weak points so that he or she can detect and supplement his or her weaknesses in advance.

Through this feedback, the system provides problems and solutions appropriate to the candidate's weak areas through an algorithm that considers the level of difficulty, so that the applicant's weak areas can be polished and their skills supplemented to help them in the next mock exam so that their effective learning It is to achieve satisfaction.

After learning the weak subject through the sixth step, the re-taking step (seventh step) of taking the test again through the test taker terminal 20 by reconnecting to the service providing server 10 is performed.

The above step is to intensively study the weak part and retake the mock test through the question bank to check the candidate's ability again.

The result of the mock test retaken in the seventh step, the retake step, is analyzed, and pass or fail is predicted, and the result prediction step, the eighth step, is performed.

Pass or fail is judged by retaking the mock test through the seventh step. Most of the qualification tests are 4 or 5 multiple-choice multiple-choice types, and if the score is 60 or higher, it is a principle to pass.

Therefore, considering that the mock test is generally more difficult than the actual qualification test, a score of 60 or less through the mock test is a failure, a pass rate of 60% for a score of 60 or more and 70 points, and a pass rate of 80% for a score of 70 or more in the main test. It can be assumed to have a probability. However, it should be noted that the above pass rate is a general example, and the percentage of pass rate can vary depending on the difficulty of the mock test.

Through the 8th step, which is the result prediction step, for example, if the test taker is expected to obtain 70 points or more and obtain a passing score of 60 points or more with a passing rate of 80% or more, the test taker can retake the mock test by his/her own choice. There is.

Therefore, through the retake of these mock exams, it is possible to secure the confidence of the test takers, check their skills, and learn a sense of practice.

In addition, even after retaking the exam, the test taker who showed a correct answer rate of 60% or less, which is also below the passing mark, receives feedback again and learns the weak part again, and retakes the mock test again.

Therefore, in the present invention, the candidate terminal 20 and the service providing server 10 communicate with each other to receive a mock test through the question bank DB 11, and through the service providing server 10 and the big data unit 30 By extracting and analyzing data on detailed items of test subjects taken by the test taker, it is possible to intensively study weak detailed items of the test taker.

When the big data unit 30 intensively provides the extracted vulnerable parts to the candidate's terminal 20, the candidate can see and learn solutions and answers to the vulnerable parts provided from the terminal 20. .

The big data unit 30 transmits the generated learning result unit 34 to the terminal 20 to feedback the learning result to the candidate, and to use the learning result to re-correct the student's learning direction, The modified direction of learning of the candidate provided to the candidate may be generated and stored as constant data. And, in the personal learning unit 35, it is possible to improve skills by finding and supplementing weak parts of each candidate.

Therefore, it is possible to provide an efficient learning method using an artificial intelligence algorithm that is suitable for the test taker and can improve learning efficiency through the continuously corrected data of the test taker's learning history.

In addition, since it is possible to predict the test taker's mock test score and accurately analyze the result of the learning level and himself, it is possible to reduce the test taker's time and cost required to improve his grades.

Hereinafter, a description will be given of a system and method for predicting a passing rate when an applicant actually takes an exam with reference to drawings.

Figure 4 is a diagram showing the configuration of the information analysis system 100, the information analysis system 100 collects data information such as information on the number of times of taking mock exams of candidates described above, mock test information, information on weak parts, etc. It is an analysis system to more accurately predict the passing rate when taking the actual exam by learning the relevance of

That is, according to the information analysis system 100, using an artificial intelligence model that has learned various information, starting from the results of the mock test, which is the most basic evaluation method, comprehensive and effective analysis of test performance analysis and pass rate prediction can be performed. can

First, the data information collected by the data collection unit 110 may include a mock test question, test take history such as the number of times taken, test difficulty, and overall ranking.

In addition, the data collection unit 110 includes actual test scoring data for each past year, including the application rate data and average score data of the previous actual exam, and mock exam scoring data for each past year, including the average score data for the previously administered mock exam, It is preferable to input and collect the mock test scoring data for the current year, including the mock test average score data for the current year, and the difficulty adjustment detail data of the actual test for each previous year.

At this time, it is explained that the questions are presented for each year of the test and the mock test, but in some cases, when the test and the mock test are conducted several times a year, each scoring data may be input and stored for each time. And, the previously administered test means the test that was conducted and scored before the date of the test for the current year on the same subject as the actual test.

The scoring data may include correct answers for each item of all applicants of the corresponding exam by year or each session of the previously administered exam. At this time, each question is classified according to subject, and score data and average score data for each subject of each candidate may be calculated according to whether each question is answered correctly and whether each question is assigned points.

In addition, pass or fail may be calculated according to the passing standard score, the passing standard score for each subject, and the like. Of course, the score for each subject, the average score, pass or fail, etc. of each candidate may be pre-calculated and provided in the scoring data.

In addition, the data collection unit 110 may input and store the data of the test take rate by past year of the actually taken test.

In the learning unit 120, among the information collected through the data collection unit 110, understanding the correlation between information such as mock test scores and overall rankings and other information (weak parts, parts that can be supplemented) AI models can be trained to do this.

In addition, it is possible to learn to understand the correlation between the mock test score and the difficulty analysis of the actual test taken in the previous year, the percentage of correct answers, and other information so that a more accurate analysis of the test is possible with only the information of the mock test score.

The input unit 130 may receive a test taker's mock test result. That is, the input unit 130 can receive characteristics (strong parts, weak parts) obtained from the test taker's test scores.

In the calculation unit 140, when the mock test score is input from the input unit 130, the artificial intelligence model learned in the learning unit 120 is used to analyze information about the weak part of the candidate and take the test. Pass rate information can be calculated.

In addition, it is preferable to calculate the learning index for each previous year by extracting the actual test scoring data for each past year and the mock test scoring data for each previous year stored in the data collection unit 110 .

In addition, the calculation unit 140 preferably calculates the learning index for the current year based on the learning index for each past year and the pre-stored data on the number of test takers for the current year.

In addition, it is preferable that the calculation unit 140 calculates the expected average score data for the current year by calculating the learning index of the current year, the average score data of the actual test of the previous year, the average score data of the mock test of the previous year, and the average score of the mock test of the current year. .

At this time, it is preferable to understand that the average score data of the actual test of the previous year is included in the scoring data of the test by the previous year, and the average score of the mock test by the previous year is included by the scoring data of the mock test by the previous year. In addition, the mock test average score data for the current year is included in the mock test scoring data for the current year.

In addition, the calculation unit 140 calculates the expected rate of taking the actual test in the current year through the average ratio of the number of test takers and the number of test takers for each previous year included in the data for the number of test takers for each previous year. this is preferable

Furthermore, the information analysis system 100 may further include a pass prediction unit 150 that predicts an actual pass rate according to the calculation of the data calculated by the calculation unit 140 .

Here, the pass prediction unit 150 automatically extracts the expected difficulty level and expected application data of the exam for the current year and transmits them to the calculation unit 140, and the calculation unit 140 firstly calculates the expected average score data for the current year. It is desirable to reflect and calculate the expected difficulty data of the exam for the current year, and calculate the final corrected expected data for the current year.

Therefore, if the candidate took the actual test in the year through the expected average score data calculated in advance based on the scoring data of the mock test before the exam for that year, the expected pass rate can be determined through the expected score.

5 is a flowchart illustrating a method for predicting a passing rate of a real test.

As shown, the step of collecting the past year and mock test scoring data of the current year (S 10) and calculating the learning index for each past year through the calculation unit 140 (S 20) and the corresponding year through the calculation unit 140 It consists of a step of calculating a learning index (S 30), a step of determining future grade information (S 40), and a step of determining an expected pass rate (S 50).

First, in the step of collecting the mock test scoring data of the past year and the current year, the test scoring data by past year including the average score data of each mock test conducted previously, the mock test scoring data by past year, and the mock test average score data of the current year Year mock test scoring data is input and stored in the data collection unit 110 (S10).

Further, the step S10 may further include inputting and storing the data on the success rate for each year of the actual test in the data collection unit 110 .

In addition, the step S10 may further include a step of calculating expected test taker rate data for the current year through each average ratio of the number of test takers and the number of test takers for each previous year included in the data on the number of test takers for each previous year. In other words, the expected number of applicants and expected number of applicants for the actual exam in the current year can be estimated through each average of the number of applicants and the number of applicants for the previous year included in the data of the number of test takers for each previous year.

It is extracted and calculated by the calculation unit 140 provided to include the mock test scoring data for each past year stored in the data collection unit 110 and the mock test scoring data for the current year and to process the calculation based on the pre-stored formula, so that the learning index for each previous year is It is calculated (S20).

Here, the learning index for each past year is calculated by [Equation 1] below.

[Equation 1]

S _t = (A _t × B _t-1 )/(A _t-1 × C _t )

Here, S _t is the learning index for each previous year in year t, A _t is the average test taker's actual test score data for each subject in year t, A _t-1 is the average test taker's actual test score data for each subject in year t-1, B _t-1 is the average mock test score data of mock test takers in year t by subject in year t-1, C _t is the average mock test score data of mock test takers in year t by subject in year t. At this time, the above-mentioned year t means the previous year excluding the current year.

And, as described above, it is preferable that the individually calculated learning index (St) for each past year be input and stored in the data collection unit 110. Subsequently, the learning index (Sn) of the current year is calculated through the calculation unit 140 based on the learning index (St) for each past year and the previously stored data on the number of test takers for each past year (S30). The learning index for the current year is as shown in [Equation 2] below.

[Equation 2]

(S _n : learning index for the current year by subject, E _t : data on the number of test takers by year t and previous year, S _t : learning index by year t and previous year)

In [Equation 2], the learning index (Sn) of the year can be calculated through the numerator value, which is the sum of the values obtained by multiplying the number of test takers by the previous year by the learning index by the previous year, and the denominator value, which is the sum of the number of test takers by past year. . And, it is preferable that the calculated learning index (Sn) for the current year is input and stored in the data collection unit 110 .

In more detail, the learning index (Sn) of the current year in [Equation 2] is the numerator value, which is the sum of the values obtained by multiplying the number of test takers by the previous year by the learning index by the previous year, and the denominator value, which is the sum of the number of test takers by past year. can be derived through And, it is preferable that the calculated learning index (Sn) for the year is input and stored in the data collection unit 110 .

Through this, since the learning index (Sn) for the current year is calculated based on the learning index (St) for each previous year and the previously stored data on the number of test takers for each previous year, it is possible to objectively determine the learning level of the test taker for the corresponding year. Therefore, since the learning index (Sn) for the current year is calculated based on the learning index (St) for each past year, the accuracy of the finally derived expected pass rate of the candidate can be improved.

In the step of determining future grade information (S40), in order to generate a score prediction model through a score prediction model generator (not shown), mock test score modeling data is collected from a plurality of test takers who have taken previous mock tests, and deep learning algorithm is used. A score prediction model is generated using the score modeling data collected based on the score.

The score prediction model generation unit constructs an artificial neural network including recurrent neural networks (RNNs), and performs neural network learning using a Backpropagation Through Time (BTPP) algorithm. The input unit 130 receives modeling data as training data. The training data includes a future score prediction of the candidate and a diagnosis result according to the learning level of the candidate.

For example, the input unit 130 may obtain the test taker's personal information according to the test subject, an actual test score within a predetermined period, a pre-measured difficulty level of the mock test, and the pre-measured mock test score as training data. Neural network learning is performed with the input unit 130, and the input unit 130 trains the candidate's personal information according to the test, the actual test score within a predetermined period, the difficulty of the pre-measured mock test, and the pre-measured mock test score. data can be obtained.

The calculation unit 140 may process it into a preset format so that the learning unit 120, which will be described later, can use at least one modeling data obtained for learning.

The data collection unit 110 may select modeling data necessary for learning from preprocessed data. For example, it may be selected according to the level according to the user's personal information or actual mock test scores, and the selected modeling data may be provided to the learning unit 120 . The learning unit 120 may learn a criterion for predicting a score using information obtained from modeling data.

For example, the learning unit 120 may learn a first criterion for which layer among a plurality of layers included in the score prediction model to apply the converted sensor data to predict the user's score. Here, the first criterion may include the type and number of evaluation data or data used to predict a score using a score prediction model, the type and level of a combination operation, and the like.

Also, when learning of the score prediction model is completed, the learning unit 120 may store the score prediction model in an arbitrary memory (not shown). Accordingly, the score prediction unit (not shown) may perform a score prediction process of the test taker using the score prediction model. In addition, the score prediction model may be stored in a memory of a server (not shown) connected to a wired or wireless network.

In addition to the future score prediction of the candidate obtained in S40 above, the learning index of the year described in the previous S30, the average test score data of the previous year, the average score data of the mock test of the previous year, and the average score of the mock test of the current year It is calculated through the calculation unit 140, and data of the expected pass rate is calculated from the pass prediction unit 150 (S50).

The calculation of the data may be based on the expected average score data (A _n ) and data on the number of test takers for the current year stored in the data collection unit 110, so that the expected pass rate of the test takers for the current year may be calculated. At this time, the expected pass rate may be calculated as a ratio of the number of successful candidates to the total number of test takers.

Therefore, it is possible to determine the expected passing rate of the exam of the current year through expected average score data calculated in advance based on the scoring data of the mock exam before the exam of the current year.

As described above, since comprehensive and accurate analysis is possible by collecting mock test scores, weak parts, etc., and learning an artificial intelligence model to understand the collected information and other information, information on mock test results can be confirmed with high accuracy. It is possible to reduce the time and effort of candidates required for diagnosis of their skills.

In addition, the information analysis system 100 can increase the analysis accuracy of mock test scores and weak parts in the process of collecting and learning information, and the accuracy of the system according to the accumulation of related data such as mock tests in the past year or actual qualification tests already taken that can be improved.

Those skilled in the art to which the present invention pertains as described above will understand that it can be implemented in other specific forms without changing the technical spirit or essential characteristics of the present invention. Therefore, it should be understood that the above-described embodiments are illustrative and not limiting.

The scope of the present invention is indicated by the appended claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts should be construed as being included in the scope of the present invention. do.

10: Service providing server 11: Question bank DB
20: test taker terminal 30: big data unit
31: learning date part 32: grades increase judgment part
33: difficulty analysis unit 34: test result storage unit
35: personal learning unit 100: information analysis system
110: data collection unit 120: learning unit
130: input unit 140: calculation unit
150: pass prediction unit

Claims (5)

Building a question bank DB including a plurality of question sets by a service providing server that provides an online evaluation service to test taker terminals; Step of selecting at least one question set to be set through the question bank DB by the service providing server; , Extracting questions from the problem set selected in the above step for each part and constructing multiple-choice options, constructing mock test questions through the service providing server, creating a mock test by mixing the order of multiple-choice questions, and In the efficient learning method of the question bank question method using an artificial intelligence algorithm comprising the step of providing a mock test to a test taker terminal,

a problem providing step of providing online a problem to be solved by a test taker accessing the service providing server;
An information receiving step of receiving information including the learner's marking content for the problem;
an information converting step of converting the information received by the information receiving step into data;
A marking item analysis step of analyzing the item marked by the candidate through the data analysis stored by the information data conversion step; and
an item analysis information providing step of providing analysis information on the marked item analyzed by the marking item analysis step to the candidate;
Learning based on weaknesses through the item analysis information providing step;
After the learning, a re-taking step of reconnecting to the service providing server and taking the test through the test taker terminal;
A result prediction step of analyzing the retake result in the retake step and predicting pass or fail,
In the learning method, when the test taker takes the test, the information analysis system 100 that can predict the passing rate is a data collection unit 110 that collects data including mock test questions, number of times of taking the test, test taking history, test difficulty, and rank. ; a learning unit 120 that obtains a correlation between a mock test score, total ranking information, a weak part, and a part to be supplemented among information collected through the data collection unit 110; An input unit 130 that receives the test taker's mock test results; a calculation unit 140 for calculating analysis information on the weak part of the candidate learned in the learning unit 120; It consists of; pass prediction unit 150 for predicting the actual pass rate,

The pass prediction method of the pass prediction unit 150 includes the step of calculating the learning index for each previous year in the calculation unit 140 (S20) and the step of calculating the learning index for the current year through the calculation unit 140 (S30) Efficient learning method of question bank question method using artificial intelligence algorithm, characterized in that it includes the step of determining future grade information (S 40) and the step of determining the expected pass rate (S 50).

The learning index for each past year is calculated by [Equation 1] below,
[Equation 1]
S _t = (A _t × B _t-1 )/(A _t-1 × C _t )
Here, S _t is the learning index for each previous year in year t, A _t is the average test taker's actual test score data for each subject in year t, A _t-1 is the average test taker's actual test score data for each subject in year t-1, B _t-1 is the average mock test score data of mock test takers in year t by subject in year t-1, C _t means the average score data of mock test takers in year t by subject in year t, and the specified year t is the previous year excluding the year means,
The learning index for the current year is as shown in [Equation 2] below.
[Equation 2]

(S _n : learning index for the current year by subject, E _t : data on the number of test takers by year t and previous year, S _t : learning index by year t and previous year)
According to claim 1,
In the marking question analysis step, if less than 30% of the entire test taker is correct, A grade difficulty, 30 to 60% less than B grade difficulty, and 60% or more are determined as C grade difficulty using an artificial intelligence algorithm, characterized in that Efficient learning method of question bank exam method.
According to claim 1 or 2,
The marking question analysis step is a question bank using an artificial intelligence algorithm, characterized in that through the detailed items of the test taker, if the test taker is all wrong, it is classified as A1, some correct and some incorrect parts as B1, and all correct parts as C1. Efficient learning method of the exam method.
According to claim 3,
The marking item analysis step provides the test taker with the parts classified as the A1, the B1, and the C1, so that the test taker learns intensively.
According to claim 1,
Efficient learning method of question bank question method using artificial intelligence algorithm, characterized in that to retake the mock test by selection if it is expected to obtain a passing score or higher in the result estimation step.

Images