KR20170034106A - Apparatus of Recommending Problems of Adequate Level for User and Method thereof - Google Patents

Apparatus of Recommending Problems of Adequate Level for User and Method thereof Download PDF

Info

Publication number
KR20170034106A
KR20170034106A KR1020150132370A KR20150132370A KR20170034106A KR 20170034106 A KR20170034106 A KR 20170034106A KR 1020150132370 A KR1020150132370 A KR 1020150132370A KR 20150132370 A KR20150132370 A KR 20150132370A KR 20170034106 A KR20170034106 A KR 20170034106A
Authority
KR
South Korea
Prior art keywords
difficulty
user
test
appropriate
degree
Prior art date
Application number
KR1020150132370A
Other languages
Korean (ko)
Inventor
손경아
박호민
이재연
정현환
Original Assignee
아주대학교산학협력단
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 아주대학교산학협력단 filed Critical 아주대학교산학협력단
Priority to KR1020150132370A priority Critical patent/KR20170034106A/en
Publication of KR20170034106A publication Critical patent/KR20170034106A/en

Links

Images

Classifications

    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B7/00—Electrically-operated teaching apparatus or devices working with questions and answers
    • G09B7/06—Electrically-operated teaching apparatus or devices working with questions and answers of the multiple-choice answer-type, i.e. where a given question is provided with a series of answers and a choice has to be made from the answers
    • G09B7/07—Electrically-operated teaching apparatus or devices working with questions and answers of the multiple-choice answer-type, i.e. where a given question is provided with a series of answers and a choice has to be made from the answers providing for individual presentation of questions to a plurality of student stations

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention relates to a device and a method for recommending an appropriate problem to a user's level.
The user's difficulty problem recommendation apparatus according to the present invention receives the test result data including the results of the users solving at least one or more problems included in the test in at least one test in which at least one user participates, A data collection and analysis unit for analyzing the test result data to determine an appropriate degree of difficulty for the user in the test in which the user participates; and a data collecting / analyzing unit for analyzing, based on the appropriate degree of difficulty for the user in the test, An inference engine unit for calculating an appropriate prediction difficulty for the user in the test not participating in the test and a problem recommendation unit for selecting the problem to be recommended to the user according to the calculated predicted difficulty level.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for recommending a user's difficulty level,

The present invention relates to a device and a method for recommending an appropriate problem to a user's level.

Systems and methods for recommending optimized content to users and development of apparatuses therefor have been made in various fields. For example, a system has been developed and used that automatically searches for and recommends multimedia content such as a movie, a moving picture or music suitable for a user, or a content such as a news or a material required by the user, and provides the retrieved and recommended information to the user. The user-customized content recommendation system analyzes user's recorded past usage information or profile information of a user or analyzes feature data of a population having characteristics similar to a user, And provides it to the user, thereby improving the convenience of the user.

That is, unlike the conventional method in which a user directly inputs information to a search system based on text, the above-described automated content search and recommendation system alleviates information search time and effort input by a user, It is possible to utilize the information as much as possible and to provide optimized information to the user.

There is also a need for a system that automatically evaluates a learner's ability using such an automated recommendation system and further recommends content for learning or testing that is most appropriate for a learner.

In this paper, we propose a system for recommending a suitable problem for users, as described in Lin et al. (LIN, S., ZHANG, Q. & E-learning, E-business, Enterprise Information Systems, and E-Government, 2013), but the existing method has limitations in manually tagging each problem, There is a problem in that it is difficult to recommend an appropriate problem to the user. In addition, in the case of the existing method, since the degree of difficulty of the problem is also set to an arbitrary criterion without reflecting the level of the user, there is a limit in that it is difficult to formulate the problem of the difficulty suitable for the user.

LIN, S., ZHANG, Q. & LI, W. (2013) A programmer self-training system with programming skill evaluation and personalized task recommendation. Proc. International Conference on E-learning, E-business, Enterprise Information Systems, and E-Government.

It is an object of the present invention to provide a method and apparatus for automatically calculating a degree of difficulty of a test when a test including a plurality of problems and a user's participation result are given thereto and analyzing a given test result data, And to provide a method and a system for selecting and recommending problems having a level of difficulty most appropriate to the level of the user.

In order to solve the above problems, a user's difficulty problem recommendation apparatus according to one type of the present invention includes a result that the users solve at least one problem included in the test in at least one test in which at least one user participates A data collection and analysis unit for receiving the test result data and analyzing the received test result data to determine an appropriate degree of difficulty for the user in the test in which the user participates; And an inference engine unit for calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the user participates.

Here, the recommendation apparatus may further include a problem recommendation unit that selects the problem to be recommended to the user according to the calculated predicted difficulty level.

Wherein the data collection and analysis unit receives the test result data generated according to the test through the network and stores the appropriate difficulty level determined for each user for the test in a database, And calculates the appropriate predicted degree of difficulty based on the received desired degree of difficulty.

Wherein the data collection and analysis unit analyzes the results of the tests in which the respective users solve the problems included in the test on the basis of the test result data and analyzes the results of the tests in accordance with the degree of difficulty of the problems, And determining the appropriate degree of difficulty for each user.

The data collection and analysis unit determines the appropriate difficulty level for each user in the test according to the degree of difficulty of the problem having the highest degree of difficulty among the problems whose results are solved for each of the users . ≪ / RTI >

Wherein the data collection and analysis unit determines the degree of difficulty according to the number of users who answered the results of the problem, and determines the degree of difficulty as the degree of the problem is high as the number of users decreases have.

Wherein the reasoning engine unit generates an appropriate difficulty matrix for each user in each of the tests using the appropriate degree of difficulty determined by the data collection and analysis unit as an element of the matrix, And calculating the predicted degree of difficulty for the user in the test not being performed.

Here, the inference engine unit predicts a component matrix obtained by decomposing the appropriate difficulty matrix using a matrix decomposition technique, calculates a prediction matrix by calculating the component matrix, and calculates the predicted difficulty according to an element value of the prediction matrix And the setting is performed.

Wherein the inference engine unit predicts the component matrix using the matrix decomposition technique considering at least one of the total difficulty level deviation, the degree of difficulty of the user, or the degree of difficulty of the test, And calculating the predicted degree of difficulty according to the prediction matrix.

Here, the inference engine unit may use the K nearest neighbor prediction method to calculate the predicted difficulty level for the user in the test in which the user is not participating, from among the other users participating in the test, K Selects the other users and calculates the predicted degree of difficulty of the user according to the appropriate degree of difficulty of the selected other users.

Here, the reasoning engine may calculate an edit distance between a column or a row corresponding to the user in the appropriate difficulty matrix and a column or a row corresponding to the other user, and use the calculated distance as the distance.

Wherein the problem recommendation unit selects the problem having the degree of difficulty within a predetermined difference from the appropriate predicted degree of difficulty among the problems stored in the database on the basis of the appropriate predicted difficulty calculated by the inference engine unit have.

According to one aspect of the present invention, there is provided a method of recommending a user's difficulty level problem, the method comprising: at least one test in which at least one user participates, Analyzing the inputted test result data and determining the appropriate degree of difficulty for the user in the test in which the user participates; An inferiority prediction step of calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the inference engine participated by the user; And the problem recommendation step may include a problem recommendation step of selecting the problem to be recommended to the user according to the calculated predicted difficulty level.

Wherein the data collection and analysis step comprises the steps of: the data collection and analysis unit analyzing the results of each of the users solving the problems included in the test by the user on the test data in the test result data; And determining the appropriate degree of difficulty for each user in the test according to the degree of difficulty.

Herein, the inferring difficulty estimating step may include a step of, in the inferring engine unit, generating a proper difficulty matrix for each of the users in each of the tests, the optimum difficulty matrix having the appropriate difficulty determined by the data collection and analysis unit as an element of the matrix, And calculating the predicted degree of difficulty for the user in the test in which the user is not participating.

According to an aspect of the present invention, there is provided a system for recommending a user's desired difficulty problem, the system including a user's desired difficulty problem recommendation server and a client apparatus.

Here, the user fairness degree problem recommendation server may receive test result data including a result that the users have solved at least one or more problems included in the test in at least one test in which at least one user participated, Analyzing the result data to determine an appropriate degree of difficulty for the user in the test in which the user participates; And an inference engine unit for calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the user participates.

Here, the client apparatus may include a problem recommendation unit for selecting the problem to be recommended to the user according to the calculated predicted difficulty level.

In order to solve the above problems, the user's desired difficulty problem recommendation program according to one type of the present invention may be a computer program stored in the medium to implement the above-described user's desired difficulty problem recommendation method.

According to the present invention, it is possible to correctly calculate and predict the appropriate level of the user, that is, the appropriate difficulty level. In addition, by recommending the most suitable problems to the user according to the degree of difficulty of the user and providing them to the user, it is possible to prevent the user from solving the problem that is too difficult or too easy when compared with the user's level, It is effective. The present invention provides an optimum difficulty problem recommendation system and an apparatus thereof. The user has the advantage of obtaining an optimal learning effect by investing in a short period of time. In view of the person who performs the test, There is an effect. There is also an effect that users' motivation for learning can be inspired more. Also, according to the present invention, it is possible to predict the appropriate difficulty level for each user in consideration of the level deviation of each competition and users.

1 is a block diagram of an apparatus for recommending a user's difficulty problem according to an embodiment of the present invention.
2 is a block diagram of an apparatus for recommending a user's difficulty problem according to another embodiment of the present invention.
3 is a block diagram of a system for recommending a user difficulty problem according to another embodiment of the present invention.
4 (a) is a reference diagram showing an example in which the test data obtained by the data collection and analysis unit 100 is expressed in the form of a two-dimensional table.
4 (b) is a reference diagram for explaining an operation of the data collection and analysis unit 100 for determining the appropriate difficulty level in each test of each user.
5 is a reference diagram showing an example of an appropriate difficulty matrix (ALM).
6 is a reference diagram showing the operation of the inference engine unit 200. As shown in FIG.
7 is a reference diagram for explaining the operation of the inference engine unit 200 when the degree of deflection is further considered.
8 is a reference diagram showing an operation of the inference engine unit 200 to select K other users near the user based on the edit distance.
9 is a flowchart of a method for recommending a user's difficulty problem according to another embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals are used to designate the same or similar components throughout the drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. In addition, the preferred embodiments of the present invention will be described below, but it is needless to say that the technical idea of the present invention is not limited thereto and can be variously modified by those skilled in the art.

Systems and methods for recommending optimized content to users and development of apparatuses therefor have been made in various fields. For example, a system has been developed and used that automatically searches for and recommends multimedia content such as a movie, a moving picture or music suitable for a user, or a content such as a news or a material required by the user, and provides the retrieved and recommended information to the user. The user-customized content recommendation system analyzes user's recorded past usage information or profile information of a user or analyzes feature data of a population having characteristics similar to a user, And provides it to the user, thereby improving the convenience of the user.

Such an automated content recommendation system is also applied to an advertisement field, and is also used in a system for displaying a customized advertisement to a user using online contents.

That is, unlike the conventional method in which a user directly inputs information to a search system based on text, the above-described automated content search and recommendation system alleviates information search time and effort input by a user, It is possible to utilize the information as much as possible and to provide optimized information to the user.

The present invention provides a system for evaluating a learner's ability using such an automatic recommendation system and further automatically recommending a content for learning or testing most suitable for a learner.

More specifically, the present invention analyzes test results of users who solve problems by participating in tests including a plurality of problems, and calculates appropriate difficulty levels of the respective tests based on the results of the tests. The present invention provides a system and an apparatus for predicting a desired degree of difficulty of a user and recommending a problem according to a predicted desired degree of difficulty to a user, and a method therefor.

In the case of the automatic test recommendation system, unlike the conventional simple recommendation system, there is data on the result of the user participating in the test, and each problem has a different difficulty from each other exist. Therefore, we can not simply apply the techniques that have been applied to the recommendation recommendation system.

In this paper, we propose a system for recommending a suitable problem for users, as described in Lin et al. (LIN, S., ZHANG, Q. & E-learning, E-business, Enterprise Information Systems, and E-Government, 2013), but the existing method has limitations in manually attaching tags to each problem, There is a problem in that it is difficult to recommend an appropriate problem to the user. In addition, in the case of the existing method, since the degree of difficulty of the problem is also set to an arbitrary criterion without reflecting the level of the user, there is a limitation that it is difficult to formulate the problem of the difficulty suitable for the user

It is an object of the present invention to provide a method and apparatus for automatically calculating a degree of difficulty of a test when a test including a plurality of problems and a user's participation result are given thereto and analyzing a given test result data, And to provide a method and a system for selecting and recommending problems having a level of difficulty most appropriate to the level of the user. In the case of analyzing the test result data, the test participants can solve the problem under the same conditions. Therefore, it is possible to measure the difficulty level of the user's ability and the problem more objectively than the conventional method.

For this purpose, the present invention automatically sets the degree of difficulty of the problem from the test result data without passive tagging and analyzes the user's test result on the problems having the set degree of difficulty automatically, As shown in FIG. Based on the adequacy extracted automatically, we predict the user's difficulty level in the tests that the user does not participate in and predict the user's problem according to the predicted level of difficulty among the problems included in the test And provides a means for recommending it to the user.

According to the present invention, it is possible to correctly calculate and predict the appropriate level of the user, that is, the appropriate difficulty level. In addition, by recommending the most suitable problems to the user according to the degree of difficulty of the user and providing them to the user, it is possible to prevent the user from solving the problem that is too difficult or too easy when compared with the user's level, It is effective. The present invention provides an optimum difficulty problem recommendation system and an apparatus thereof. The user has the advantage of obtaining an optimal learning effect by investing in a short period of time. In view of the person who performs the test, There is an effect. There is also an effect that users' motivation for learning can be inspired more.

Hereinafter, an apparatus for recommending a user's difficulty problem according to the present invention and a method thereof will be described in detail.

1 is a block diagram of an apparatus for recommending a user's difficulty problem according to an embodiment of the present invention.

The user adequacy problem recommendation apparatus according to the present invention may include a data collection and analysis unit 100 and an inference engine unit 200 and may further include a problem recommendation unit 300 as needed.

Here, the apparatus for recommending a user difficulty problem recommendation apparatus according to the present invention may be implemented as a computer program having a program module that performs a part or all of the functions of a part or all of the constituent elements selectively combined and combined in one or a plurality of hardware It is possible. In addition, each component may be implemented as a single independent hardware or included in each hardware as needed. Also, the apparatus for recommending a user difficulty problem recommending apparatus according to the present invention may be implemented as a software program and operated on a processor or a signal processing module, or may be implemented in hardware to be included in various processors, chips, semiconductors, have. Further, the user's difficulty problem recommendation apparatus according to the present invention may be included in the form of hardware or software modules on a computer, various embedded systems or devices. Preferably, the user's difficulty problem recommendation apparatus according to the present invention may be implemented in a server connected to a network or included in a server. The data collection and analysis unit 100, the inference engine unit 200, and the problem recommendation unit 300 of the user's optimum difficulty problem recommendation apparatus according to the present invention may be implemented on one user recommended difficulty problem recommendation server, And some configurations may be implemented on different servers or may exist on a plurality of servers as needed. In addition, some configurations may be implemented in the client device or included in the client device, but not in the server as needed. For example, the data collection and analysis unit 100 and the reasoning engine unit 200 may be included in the server, and the problem recommendation unit 300 may be included in the client apparatus.

The data collection and analysis unit 100 receives the test result data including the results of the users solving at least one or more problems included in the test in at least one test in which at least one user participated, And determines an appropriate degree of difficulty for the user in the test in which the user participates.

The inference engine unit 200 calculates an appropriate predicted degree of difficulty for the user in the test in which the user is not participating based on the appropriate degree of difficulty for the user in the test in which the user participates.

The problem recommendation unit 300 selects the problem to be recommended to the user according to the calculated predicted difficulty level.

2 is a block diagram of an apparatus for recommending a user's difficulty problem according to another embodiment of the present invention.

The data collection and analysis unit 100, the reasoning engine unit 200, and the problem recommendation unit 300 of the apparatus for recommending a user's desired difficulty problem according to another embodiment of the present invention may operate in connection with the database 20. For example, the server 10 as a user difficulty problem recommendation apparatus may include a data collection and analysis unit 100, an inference engine unit 200, a problem recommendation unit 300, and a user suitability difficulty problem recommendation server 10 And the database 20 may be connected to operate through a network.

Here, the data collection and analysis unit 100 receives the test result data generated according to the test through the network, and stores the appropriate difficulty levels determined for each user in the database in the test.

At this time, the inference engine unit 200 receives the appropriate difficulty level from the database 20 and can calculate the appropriate predicted difficulty level based on the received optimal difficulty level.

In this case, the problem recommendation unit 300 may further include, among the problems stored in the database 20 on the basis of the appropriate predicted difficulty calculated by the inference engine unit 200, You can select the problem.

3 is a block diagram of a system for recommending a user difficulty problem according to another embodiment of the present invention.

In the user desired difficulty problem recommendation system according to another embodiment of the present invention, the user desired difficulty problem recommendation server 10 may include a data collection and analysis unit 100 and an inference engine unit 200, And may include a problem recommendation unit 300.

That is, the user fairness degree problem recommendation system 1 may include a user fairness degree problem recommendation server 10 and a client device 30. At this time, the user fairness degree problem recommendation server 10 may include a user fairness degree problem recommendation server 10, The test result data including the results of the users solving at least one or more problems included in the test in at least one of the tests, analyzing the inputted test result data, Based on the appropriate degree of difficulty for the user in the test in which the user participates in the test, the data collection and analysis unit 100 for determining the appropriate difficulty level for the user, And an inference engine unit 200 for calculating an inference engine. At this time, the client device 30 may include a problem recommendation unit 300 for selecting the problem to be recommended to the user according to the calculated predicted difficulty level.

At this time, the user fairness degree problem recommendation server 10, the database 20, and the client device 30 may be connected to each other via a network to operate while transmitting and receiving information. For example, when the problem recommendation unit 300 exists in a different apparatus from the inference engine unit 200 as shown in FIG. 3, the predicted difficulty level can be received and used through the network.

Next, the operation of the data collection and analysis unit 100 will be described in more detail.

The data collection and analysis unit 100 may receive the test result data first. Wherein the test result data may include a result of the users solving at least one or more problems included in the test in at least one or more tests in which at least one user participated. That is, the test result data may include the results obtained by the users who participated in each test in each test to solve the problems included in the test. And the result of solving the problem here can be information indicating whether or not it succeeded in solving the problem or whether the correct answer of the problem was met.

For example, if the user succeeds in solving a specific problem, the result can be 1, and if the user fails to solve a particular problem, the result of solving the problem can be zero. Here, 1, 0 is an example of expressing a result value, and it can be stored in test result data by using various characters or symbols (O, X).

4 (a) is a reference diagram showing an example of expressing the test result data inputted by the data collection and analysis unit 100 in the form of a two-dimensional table. 4A, the test result data includes data (U1, U2, U3) participating in the specific test (Contest 1) including the results of solving the problems (P1, ... P5) , And the resultant value can be represented by O, X.

The data collection and analysis unit 100 receives the test result data and stores the data in the form of an array or other data structure. Alternatively, test result data stored in the form of a constant data structure may be input.

Next, the data collection and analysis unit 100 analyzes the test result data inputted as described above to determine the appropriate degree of difficulty for the user in the test in which the user participates. Here, the appropriate degree of difficulty refers to an index indicating the level of the user in the test. In the present invention, the index indicating the level of the user is determined as a value determined according to the degree of difficulty of the user who answers the questions as a result of the user solving the problems included in the test.

Therefore, the data collection and analysis unit 100 analyzes the results of the above-mentioned problems included in the test by each user in the test result data on the basis of the test results, The appropriate degree of difficulty for each user in the test can be determined.

Here, the difficulty of each problem may be a preset value for each problem. More preferably, however, the difficulty of each problem can be determined by the number of users who have succeeded in solving each problem, i.e., the number of users who have matched the correct answer to the problem. That is, the data collection and analysis unit 100 preferably sets the difficulty value of each problem according to the number of users who have succeeded in solving each problem.

The data collection and analysis unit 100 determines the degree of difficulty based on the number of users who answered the results of the problem, and determines the degree of difficulty as the number of users is high, Do. In other words, if the degree of difficulty is defined as having a higher numerical value, the data collection and analysis unit 100 determines that the larger the number of users who succeed in solving each problem, the smaller the difficulty value and the smaller the number of users succeeding in solving each problem It is desirable to set the difficulty level of the problem so that the recorded difficulty value becomes larger. For example, the data collection and analysis unit 100 may calculate the degree of difficulty of a specific problem so as to be proportional to the reciprocal of the number of users who matched the problem.

At this time, the data collection and analysis unit 100 may calculate the appropriate difficulty level for each user in the test according to the degree of difficulty of the predetermined difficulty among the problems, Can be determined. That is, the data collecting and analyzing unit 100 may calculate the appropriate difficulty level for each user in the test according to the difficulty value of the problems at a predetermined ratio of the difficulty level among the difficulty levels of the problems, . For example, the data collection and analysis unit 100 can determine the appropriate degree of difficulty of the user according to the difficulty of the problems having the difficulty of the top 10% among the problems that the user has answered correctly. For example, The average value of the difficulty of the problems, or the median value may be determined by the appropriate difficulty level.

In this case, the data collection and analysis unit 100 preferably stores the difficulty of the problem having the highest degree of difficulty among the problems, which is the result of solving the problem for each user, It can be decided by the degree of difficulty. In other words, it is desirable that the appropriate degree of difficulty is the degree of difficulty of the problem which is the most difficult among the users.

FIG. 4 (b) is a reference diagram showing the result of determining the degree of difficulty of each user in the corresponding test with the degree of difficulty having the highest degree of difficulty among users who have successfully solved the test.

The data collection and analysis unit 100 may collect appropriate difficulty levels determined for each user in each test as described above and store the collected difficulty levels in a storage unit such as a memory. The appropriate difficulty level data collected in this manner can be stored in the database 20.

Next, the operation of the inference engine unit 200 according to the present invention will be described in more detail.

The inference engine unit 200 calculates an appropriate predicted degree of difficulty for the user in the test in which the user is not participating based on the appropriate degree of difficulty for the user in the test in which the user participates.

That is, since there exists a test in which the user solves the problem and has not yet participated in the test, the data collecting and analyzing unit 100 analyzes the result data of the test in which the user participates, Based on the appropriate degree of difficulty, the reasoning engine unit 200 predicts the appropriate difficulty level of the user in the test in which the user has not yet participated and has not solved the problem. Thus, the predicted value of the user's desired degree of difficulty in the test in which the user has not yet participated is called the predicted degree of difficulty.

The inference engine unit 200 first generates an appropriate difficulty matrix for each user in each of the tests, with the appropriate degree of difficulty determined by the data collection and analysis unit 100 as an element of the matrix.

Wherein the appropriate difficulty matrix has an appropriate difficulty value for the user in the test in which the user participates as an element value of the matrix and an appropriate unknown value for the user in the test in which the user is not participating The difficulty value can be an element of the matrix as an unknown number.

FIG. 5 is a reference diagram showing an example of such an appropriate difficulty matrix (ALM). (U1, ..., Un) participates in m tests (Contest) up to C1, ..., Cm. In each test, the data collecting and analyzing unit (100) The calculated desired difficulty levels can be integrated and expressed as a proper difficulty matrix as shown in FIG. At this time, each element value of the appropriate difficulty matrix becomes the appropriate difficulty value of the corresponding user in the corresponding test. At this time, since the data collection and analysis unit 100 does not calculate the appropriate difficulty value for the user who has not participated in the specific test, this is left as an unknown number. In FIG.

The inference engine unit 200 calculates the appropriate prediction difficulty for the user in the test in which the user is not participating using the appropriate difficulty matrix. That is, the inference engine unit 200 predicts the unknowns represented by '-' symbols in FIG. 5 using the appropriate difficulty values already known in the appropriate difficulty matrix.

Here, in order to calculate the predicted difficulty level for the user in the test in which the user is not participating, in other words, in order to predict the values of the elements whose values are left as unknowns in the appropriate difficulty matrix, By using the appropriate difficulty level obtained from the users' test results as input, it is possible to predict the appropriate difficulty level of the unknown using various existing prediction methods.

At this time, the inference engine unit 200 may include, for example, an existing 'Corinna Cortes. and Vladimir Vapnik., "Support-vector networks ", Machine Learning, Vol. 20, Issue 3, pp. 273-297, 1995. 'Support Vector Machine (SVM) technique', David A. Freedman (2009). Statistical Models: Theory and Practice. Cambridge University Press. p. 128. The Logistic Regression Method or the Generalized Linear Model, etc., can be used to predict the remaining difficulty levels as unknowns in the appropriate difficulty matrix.

However, there is a high possibility that the number of users participating in the test is not relatively large, and accordingly, the appropriate difficulty matrix is likely to be a matrix having many unknown elements. That is, the appropriate difficulty matrix has a high probability of having a sparsity.

And it can be said that the prediction method based on collaborative filtering is desirable for reliable prediction from the data with high scarcity. Therefore, the inference engine unit 200 according to the present invention performs prediction using a matrix factorization method or a K Nearest Neighborhood method. At this time, the inference engine unit 200 is' Koren, Yehuda, Robert Bell, and Chris Volinsky. "Matrix factorization techniques for recommender systems." Computer 42.8 (2009): 30-37 ', and' Jannach, Dietmar, et al. Recommender systems: an introduction. Cambridge University Press, 2010. p13-18 '.

Particularly, for the problem of predicting the appropriate degree of difficulty of a specific user who does not participate in a specific test from the test data extracted from the test results of users participating in the test, which is a problem to be solved by the present invention, It is experimentally confirmed that the predicted difficulty can be calculated more precisely as described below. Therefore, most preferably, the inference engine unit 200 can perform prediction using a matrix factorization technique. Also, it has been experimentally confirmed that an excellent prediction result is obtained even in the case of using the K nearest neighbor method. It is preferable that the inference engine unit 200 may perform prediction using the K nearest neighbor method.

First, the reasoning engine 200 predicts the appropriate predictive difficulty by applying a matrix decomposition technique to solve the problems of the present invention.

The inference engine unit 200 predicts a component matrix obtained by decomposing the appropriate difficulty matrix using a matrix decomposition technique, calculates a prediction matrix by calculating the component matrix, and calculates an optimal prediction matrix based on the element value of the prediction matrix. It is desirable to set the difficulty level.

Herein, the inference engine unit 200 uses a matrix decomposition technique to calculate a given fairness degree matrix F by using a first component matrix P representing information according to the user and a second component matrix Q representing information according to the test And the values of the elements of the prediction matrix F 'calculated by multiplying the first component matrix P and the second component matrix Q thus decomposed are compared with the values of the elements of the appropriate difficulty matrix, And predicts the first component matrix P and the second component matrix Q so that the difference becomes small. The elements of the appropriate difficulty matrix F are left as unknowns according to the element values of the prediction matrix F 'finally calculated by multiplying the predicted first component matrix P and the second component matrix Q Set the appropriate predictability level of the elements.

The appropriate difficulty matrix F may be decomposed into a first component matrix P representing information according to the user and a second component matrix Q representing information according to the test as shown in Equation 1 below.

Figure pat00001

The appropriate difficulty matrix F may be a matrix of N x M, P may be a matrix of N x K, and Q may be a matrix of K x M.

And the prediction matrix F 'can be calculated as a product of a matrix as shown in the following equation (2).

Figure pat00002

(M, n) , the row vector (1 x K) of the n-th row of the first component matrix P is denoted by p n , and the elements of the second component matrix Q Assuming that a column vector (K x 1) of m columns is q m , each element of the prediction matrix F 'can be expressed as a product of a vector as shown in the following Equation (3).

Figure pat00003

At this time, the inference engine unit 200 preferably calculates and predicts the first component matrix P and the second component matrix Q so as to minimize the cost function as shown in Equation (4).

Figure pat00004

Where f (n,m) ∈ F means of the elements whose value is obtained by calculating a fair degree of difficulty in collecting data analysis unit 100 at an appropriate difficulty level matrix, and D (f (n,m), f '(n , m ) ) denotes a function for calculating the distance between f ( n, m ) and f ' (n, m) . For example, D (f, f ') may be an arithmetic function such as (f - f') 2 , may be computed as | f - f '|, have. Here, λ 1 and λ 2 are regularization parameters for adjusting the degree of shaping of the respective elements.

Then, the inference engine unit 200 calculates the final prediction matrix as a product of the first component matrix and the second component matrix predicted as described above and calculates the values of the elements remaining as unknowns in the appropriate difficulty matrix by the corresponding position As shown in FIG. In other words, when an element of n rows and m columns remains as an unknown number in the desired difficulty matrix, the value of the corresponding element can be calculated from the values of the first component matrix and the second component matrix, as shown in Equation (3).

6 is a reference diagram showing the operation of the inference engine unit 200. FIG.

As shown in FIG. 6, when the inference engine unit 200 predicts the first component matrix P and the second component matrix Q obtained by decomposing the appropriate difficulty matrix F in the same manner as described above, and calculates the prediction matrix F ' The engine unit 200 can set the appropriate predicted difficulty value by setting the values of the elements that remain unknown in the appropriate difficulty matrix F by using the element values of the calculated prediction matrix F '. For example, in the third test C3, the element value of the appropriate difficulty matrix (1, 3) in which the first user U1 does not participate and the appropriate difficulty level remained unknown is set to the element value 2.1 of the prediction matrix (1, 3) , The appropriate prediction difficulty value of the first user U1 can be set in the third test C3.

At this time, the inference engine unit 200 predicts the component matrix using the matrix decomposition technique, considering at least any one of the total difficulty level deviation, the degree of difficulty of the user, or the degree of difficulty of the test , The prediction matrix may be calculated, and the appropriate prediction difficulty may be set according to the prediction matrix.

That is, the difficulty of the problems included in the test may be different from the other tests according to the tendency of questioning in each test, and likewise, the average degree of the appropriate difficulty may be different according to each user's ability. The inference engine unit 200 may perform the process of calculating the appropriate predicted degree of difficulty by further considering the degree of difficulty of the user or the degree of difficulty of the test in order to consider the bias in accordance with the test or the characteristic of the user . At this time, the overall difficulty bias considering both the test and the user may also be considered as needed.

When the total degree of difficulty is represented by μ, the degree of user's degree of difficulty is denoted by bu, and the degree of difficulty of the test is denoted by cu, Equation (3) can be changed to Equation (5) And Equation (4) for predicting the second component matrix may be modified as shown in Equation (6).

Figure pat00005

Here, the user's degree of difficulty may be a vector having an element N of the number of users, and the degree of difficulty of the test, bc, may be a vector having an element M of the number of tests. And where bc m is the deflection of the mth test of bc, and bu n is the deflection of the nth user of bu. Where μ denotes the average of the elements for which a given value has been obtained in the appropriate difficulty matrix (F). That is, μ is an average of the elements of the appropriate difficulty matrix corresponding to the appropriate degree of difficulty obtained by analyzing the data of the test performed by the user by the data collection and analysis unit 100.

Figure pat00006

Where f (n,m) ∈ F is n by calculating the appropriate difficulty level means the element with the value received from the second user in the m-th test data collection and analysis part 100 at an appropriate difficulty level matrix for, and D (f ( n, m), f '( n, m)) is f (n, m) and f' it refers to a function for calculating a distance between the (n, m). For example, D (f, f ') may be an arithmetic function such as (f - f') 2 , may be computed as | f - f '|, have. Here, λ 1 , λ 2 , λ 3 , and λ 4 are regularization parameters for adjusting the degree of shaping of each element.

7 is a reference diagram for explaining the operation of the inference engine unit 200 when the degree of deflection is further considered. The inference engine unit 200 can calculate the prediction matrix as shown in FIG. 7 (c) by decomposing the appropriate difficulty matrix as shown in FIG. 7 (a) into the matrix of (b) Can be obtained.

Next, in order to solve the problem of the present invention, the inference engine unit 200 predicts the optimal predicted difficulty using the K nearest neighbor prediction technique will be described in detail.

The nearest neighbors prediction method is a technique for predicting the class of predicted data by selecting K data close to the predicted data and then classifying the data into a class having many K data.

Herein, the inference engine unit 200 may use the K nearest neighbor prediction technique to calculate the appropriate prediction difficulty for the user in the test in which the user is not participating, It is possible to select K other users who are close to the user and to calculate the predicted degree of difficulty of the user according to the appropriate degree of difficulty of the selected other users.

For example, the reasoning engine unit 200 may set the appropriate difficulty level, which occurs most frequently in the predetermined difficulty level of the selected other users, to the appropriate predictive difficulty level. For example, when K is 4, the inference engine unit 200 can select four different users whose distances from the user are close to each other, and if the appropriate difficulty levels of the selected users are 3, 3, 3, and 5 The appropriate predicted degree of difficulty for the user can be set to 3.

At this time, the inference engine unit 200 may calculate the edit distance between the column or row corresponding to the user in the appropriate difficulty matrix and the column or row corresponding to the other user, and may make the distance.

Here, the edit distance is a distance between two strings, which is a distance calculated according to the degree to which a string must be edited in order to make different strings identical to each other. Assuming that the lengths of the two strings are the same, the edit distance can be calculated in such a way that the distance increases by 1 every time the different characters are changed. For example, to make 'abcd' and 'abdf' identical, 'cd' should be replaced with 'df', so the editing distance is 2.

When calculating the distance between the user who wants to calculate the appropriate predictive difficulty and the other user, the inference engine unit 200 calculates a vector representing a proper difficulty in each test of the user and a vector representing a proper difficulty in each test of the other user It is possible to calculate the edit distance between the two points. Here, when calculating the edit distance between different vectors, when the element values included in the vector are different, the sum of the magnitudes of the differences of the different elements can be calculated as the edit distance.

In this case, the vector representing the appropriate difficulty level for the tests for each user may be each row vector in the appropriate difficulty matrix, and if so, the inference engine unit 200 may calculate the row vector in each of the appropriate difficulty matrices of the user and the other user It is possible to obtain the distance by calculating the edit distance between the two.

Herein, the reasoning engine unit 200 does not calculate the appropriate difficulty by the data collection and analysis unit 100 in calculating the edit distance between both vectors, and compares the part including the element remaining as an unknown number, . For example, the edit distance between 'abcd' and 'abc-' can be calculated to be zero.

Herein, the inference engine unit 200 may calculate the edit distance as shown in Equation (7).

Figure pat00007

Here, a and b are different row vectors in the appropriate difficulty matrix, '-' is a symbol representing an element remaining as the unknown number, and W is the edit distance.

FIG. 8 is a reference diagram showing an operation of the inference engine unit 200 to select K other users near the user based on the edit distance. The inference engine unit 200 can calculate the edit distance between the row vectors as shown in FIG. 8A, and can select K other users whose distance from the specific user is close to the specific user as shown in FIG. 8B .

Next, the operation of the problem recommendation unit 300 according to the present invention will be described in more detail.

The problem recommendation unit 300 selects the problem to be recommended to the user according to the calculated predicted difficulty level.

The problem recommendation unit 300 selects the problem predicted difficulty calculated by the inference engine unit 200 from among the problems stored in the database 20 based on the appropriate predicted difficulty, Can be selected. For example, when the appropriate predicted difficulty level is calculated as 3, the problem recommendation unit can select problems having a difficulty level of 3 in the database 20, or problems with a difficulty level of 3 or less, that is, difficulty levels of 3, 4, 5). In this way, the problem recommendation unit 300 can select problems to be recommended to the user, with difficulty in the range appropriately selected by the administrator or the user, as needed, based on the estimated predicted difficulty calculated by the inference engine unit 200 . This is true even when the appropriate predicted difficulty is calculated as a prime rather than an integer.

9 is a flowchart of a method for recommending a user's difficulty problem according to another embodiment of the present invention.

The method of recommending a user's desired degree of difficulty according to another embodiment of the present invention may include a data collection analysis step (S100), an appropriate difficulty level prediction step (S200), and a problem recommendation step (S300). Hereinafter, the method for recommending a user's desired difficulty problem according to the present invention will be described in detail with reference to the data collection and analysis unit 100, the reasoning engine unit 200, the problem recommendation unit 300, Can be operated in the same manner as < / RTI > The overlapping description will be omitted and briefly described. Each operation of the data collection analysis step (S100), the appropriate difficulty level prediction step (S200), and the problem recommendation step (S300) will be described in more detail with reference to the data collection analysis part The inference engine unit 200, and the problem recommendation unit 300, as shown in FIG.

In the data collection analysis step S100, the data collection and analysis unit 100 inputs test result data including the results of the users solving at least one or more problems included in the test in at least one test in which at least one user participated And analyzes the inputted test result data to determine the appropriate degree of difficulty for the user in the test in which the user participates.

In the appropriate difficulty level prediction step S200, the inference engine unit 200 determines an appropriate prediction level for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the user participated It calculates the difficulty level.

In the problem recommendation step S300, the problem recommendation unit 300 selects the problem to be recommended to the user according to the calculated predicted difficulty level.

Here, in the data collection and analysis step (S100), the data collection and analysis unit (100) analyzes the results of each of the users who have solved the problems included in the test by the test on the test result data, The appropriate degree of difficulty for each user in the test can be determined according to the degree of difficulty of the correct answers.

Here, in the proper difficulty predicting step S200, the inference engine unit 200 calculates a proper difficulty matrix for each user in each of the tests by using the appropriate difficulty determined in the data collection and analysis unit 100 as an element of the matrix And calculate the appropriate prediction difficulty for the user in the test in which the user is not participating using the appropriate difficulty matrix.

Yet another embodiment of the present invention may be a computer program stored on a medium for implementing the method of recommending a user's difficulty problem described above in detail in connection with a computer. The computer includes a computer, a system, an embedded device, and a device in which the data collection and analysis unit 100, the reasoning engine unit 200, and the problem recommendation unit 300 are implemented or include the above configuration.

It is to be understood that the present invention is not limited to these embodiments, and all elements constituting the embodiment of the present invention described above are described as being combined or operated in one operation. That is, within the scope of the present invention, all of the components may be selectively coupled to one or more of them.

In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. In addition, such a computer program may be stored in a computer readable medium such as a USB memory, a CD disk, a flash memory, etc., and read and executed by a computer to implement an embodiment of the present invention. As the recording medium of the computer program, a magnetic recording medium, an optical recording medium, a carrier wave medium, and the like can be included.

Furthermore, all terms including technical or scientific terms have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined in the Detailed Description. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

It will be apparent to those skilled in the art that various modifications, substitutions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims. will be. Therefore, the embodiments disclosed in the present invention and the accompanying drawings are intended to illustrate and not to limit the technical spirit of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments and the accompanying drawings . The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

100: Data collection and analysis unit
200: inference engine section
300: Problem Recommendation Department
10: Recommended user difficulty problem recommendation device
20: Database
30: Client device
S100: Data collection analysis phase
S200: Estimation of appropriate difficulty level
S300: Problem Recommendation Step

Claims (16)

Wherein the test result data includes at least one or more tests in which at least one user has participated in at least one of the tests and at least one of the users has solved at least one problem included in the test, A data collection and analysis unit for determining an appropriate degree of difficulty for the user in the test; And
And an inference engine unit for calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the user participated, Appropriate difficulty problem recommendation device. The method according to claim 1,
And a problem recommendation unit for selecting the problem to be recommended to the user according to the calculated predicted difficulty level. The method according to claim 1,
Wherein the data collection and analysis unit receives the test result data generated according to the test through a network and stores the appropriate difficulty level determined for each user for the test in a database,
Wherein the inference engine unit receives the appropriate difficulty level from the database and calculates the appropriate predictive difficulty level based on the received optimal difficulty level. The method according to claim 1,
Wherein the data collection and analysis unit analyzes the results of the tests in which the respective users solve the problems included in the test on the basis of the test results data to determine whether or not the results of the tests are correct according to the degree of difficulty of the problems, And determines the appropriate degree of difficulty for the user. 5. The method of claim 4,
The data collection and analysis unit may determine the appropriate degree of difficulty for each user in the test according to the degree of difficulty of the predetermined problem having the highest degree of difficulty among the problems whose results are solved for each of the users A user-desired difficulty problem recommendation device. 5. The method of claim 4,
Wherein the data collection and analysis unit determines the degree of difficulty based on the number of users who answered the results of the problem, and determines the degree of difficulty as the degree of difficulty of the problem is higher as the number of users decreases, Appropriate difficulty problem recommendation device. The apparatus according to claim 1,
For each of the users in each of the tests, generates a proper difficulty matrix having the appropriate degree of difficulty determined by the data collection and analysis unit as an element of the matrix,
And calculates the appropriate predictive difficulty for the user in the test in which the user is not participating, using the appropriate difficulty matrix. 8. The method of claim 7,
Wherein the inference engine unit predicts a component matrix obtained by decomposing the appropriate difficulty matrix using a matrix decomposition technique, calculates a prediction matrix by calculating the component matrix, and sets the optimal prediction difficulty according to an element value of the prediction matrix Wherein the recommendation degree of the user is determined based on the user's preference. 9. The method of claim 8,
Wherein the inference engine unit predicts the component matrix using the matrix decomposition technique considering at least one of the total difficulty level deviation, the degree of difficulty of the user, or the degree of difficulty of the test, And sets the predicted degree of difficulty according to the prediction matrix. 8. The method of claim 7,
Wherein the inference engine unit compares the estimated predictive difficulty for the user in the test in which the user is not participating by using K nearest neighbor prediction techniques to K candidates, which are close to the user, among other users participating in the test Selects the other users, and calculates the predicted degree of difficulty of the user according to the appropriate degree of difficulty of the selected other users. 11. The method of claim 10,
Wherein the inference engine unit calculates an editing distance between a column or a row corresponding to the user in the appropriate difficulty matrix and a column or a row corresponding to the other user to set the distance as the distance. The method according to claim 1,
Wherein the problem recommendation unit selects the problem having the degree of difficulty within a predetermined difference from the appropriate predicted degree of difficulty among the problems stored in the database based on the appropriate predicted difficulty calculated by the inference engine unit, Appropriate difficulty problem recommendation device. Wherein the data collection and analysis unit receives test result data including at least one or more users having participated in at least one test in which at least one or more users participated in the test have solved at least one or more problems included in the test, A data collection analysis step of determining an appropriate degree of difficulty for the user in the test in which the user participates;
An inferiority prediction step of calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the inference engine participated by the user; And
Wherein the problem recommendation step includes a problem recommendation step of selecting the problem to be recommended to the user according to the calculated predicted difficulty level. 14. The method of claim 13,
Wherein the data collection and analysis unit analyzes the results of each of the users solving the problems included in the test for each of the tests on the test result data to determine whether the result of the solving the problem is correct, And determining the appropriate degree of difficulty for the user. 14. The method of claim 13, wherein the step of predicting the desired degree of difficulty comprises:
Wherein the inference engine unit generates an appropriate difficulty matrix for each of the users in each of the tests using the appropriate degree of difficulty determined by the data collection and analysis unit as an element of the matrix, And calculating the predicted degree of difficulty for the user in the test. A user-desired difficulty problem recommendation system comprising a recommendation server and a client apparatus,
The user adequacy problem recommendation server includes:
Wherein the test result data includes at least one or more tests in which at least one user has participated in at least one of the tests and at least one of the users has solved at least one problem included in the test, A data collection and analysis unit for determining an appropriate degree of difficulty for the user in the test; And
And an inference engine unit for calculating an appropriate predicted degree of difficulty for the user in the test in which the user is not participating, based on the appropriate degree of difficulty for the user in the test in which the user participated,
The client device comprising:
And a problem recommendation unit for selecting the problem to be recommended to the user according to the calculated predicted difficulty level.
KR1020150132370A 2015-09-18 2015-09-18 Apparatus of Recommending Problems of Adequate Level for User and Method thereof KR20170034106A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020150132370A KR20170034106A (en) 2015-09-18 2015-09-18 Apparatus of Recommending Problems of Adequate Level for User and Method thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020150132370A KR20170034106A (en) 2015-09-18 2015-09-18 Apparatus of Recommending Problems of Adequate Level for User and Method thereof

Publications (1)

Publication Number Publication Date
KR20170034106A true KR20170034106A (en) 2017-03-28

Family

ID=58495773

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020150132370A KR20170034106A (en) 2015-09-18 2015-09-18 Apparatus of Recommending Problems of Adequate Level for User and Method thereof

Country Status (1)

Country Link
KR (1) KR20170034106A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20190082453A (en) * 2018-01-02 2019-07-10 (주)뤼이드 Method, apparatus and computer program for analyzing new learning contents for machine learning modeling
WO2020080826A1 (en) * 2018-10-16 2020-04-23 주식회사 뤼이드 Machine learning method, device, and computer program for providing personalized educational contents on basis of learning efficiency
KR20200095655A (en) 2019-02-01 2020-08-11 주식회사 메디오피아테크 Method for recommending question based on similarity and server using the same
CN111708951A (en) * 2020-06-23 2020-09-25 广东讯飞启明科技发展有限公司 Test question recommendation method and device
KR20220056748A (en) * 2020-10-28 2022-05-06 머니매그넷 주식회사 Method and device for classifying and adjusting difficulty of evaluation problems for measuring learning level of learners level based on machine learning

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
(논문자료 001) LIN, S., ZHANG, Q. & LI, W. (2013) A programmer self-training system with programming skill evaluation and personalized task recommendation. Proc. International Conference on E-learning, E-business, Enterprise Information Systems, and E-Government (EEE).

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20190082453A (en) * 2018-01-02 2019-07-10 (주)뤼이드 Method, apparatus and computer program for analyzing new learning contents for machine learning modeling
KR20200084816A (en) * 2018-01-02 2020-07-13 (주)뤼이드 Method, apparatus and computer program for analyzing new contents for solving cold start
WO2020080826A1 (en) * 2018-10-16 2020-04-23 주식회사 뤼이드 Machine learning method, device, and computer program for providing personalized educational contents on basis of learning efficiency
JP2021508065A (en) * 2018-10-16 2021-02-25 ルイド インコーポレイテッド Machine learning methods, devices and computer programs for providing personalized educational content based on learning efficiency
US11704578B2 (en) 2018-10-16 2023-07-18 Riiid Inc. Machine learning method, apparatus, and computer program for providing personalized educational content based on learning efficiency
KR20200095655A (en) 2019-02-01 2020-08-11 주식회사 메디오피아테크 Method for recommending question based on similarity and server using the same
KR20210101201A (en) 2019-02-01 2021-08-18 주식회사 메디오피아테크 Method for recommending question based on similarity and server using the same
CN111708951A (en) * 2020-06-23 2020-09-25 广东讯飞启明科技发展有限公司 Test question recommendation method and device
CN111708951B (en) * 2020-06-23 2023-06-09 广东讯飞启明科技发展有限公司 Test question recommending method and device
KR20220056748A (en) * 2020-10-28 2022-05-06 머니매그넷 주식회사 Method and device for classifying and adjusting difficulty of evaluation problems for measuring learning level of learners level based on machine learning

Similar Documents

Publication Publication Date Title
Bürkner et al. Approximate leave-future-out cross-validation for Bayesian time series models
US10242345B2 (en) Automatic interview question recommendation and analysis
Winand et al. A unified model of non-profit sport organizations performance: perspectives from the literature
US20180365229A1 (en) Systems and methods to determine and utilize semantic relatedness between multiple natural language sources to determine strengths and weaknesses
KR20170034106A (en) Apparatus of Recommending Problems of Adequate Level for User and Method thereof
Lempert Agent-based modeling as organizational and public policy simulators
US10042944B2 (en) Suggested keywords
Chaâbane A novel auto-regressive fractionally integrated moving average–least-squares support vector machine model for electricity spot prices prediction
Evans et al. Computation and the sociological imagination
Mejri et al. An ensemble method for concept drift in nonstationary environment
MirHassani et al. Quantum binary particle swarm optimization-based algorithm for solving a class of bi-level competitive facility location problems
Salimans Variable selection and functional form uncertainty in cross-country growth regressions
Kang et al. A Bayesian hierarchical model for quantitative and qualitative responses
Kumar et al. Data, machine learning, and human domain experts: none is better than their collaboration
Koh et al. A model comparison algorithm for increased forecast accuracy of dengue fever incidence in Singapore and the auxiliary role of total precipitation information
CN113342942A (en) Corpus automatic acquisition method and device, computer equipment and storage medium
Dolnicar et al. Methods in segmentation
Aziz et al. Identifying new podcasts with high general appeal using a pure exploration infinitely-armed bandit strategy
Perälä et al. Detecting regime shifts in fish stock dynamics
Zheng et al. A doubly stochastic simulator with applications in arrivals modeling and simulation
Shao et al. A Combinatorial optimization framework for scoring students in University Admissions
Cen et al. NIM: Generative Neural Networks for Automated Modeling and Generation of Simulation Inputs
KR101981616B1 (en) Method of providing services that predict the performance of influencer marketing
Chu et al. SimCURL: Simple Contrastive User Representation Learning from Command Sequences
Kuruwita Variable selection in the single-index quantile regression model with high-dimensional covariates

Legal Events

Date Code Title Description
A201 Request for examination
E902 Notification of reason for refusal
E601 Decision to refuse application