KR20220170654A - System of learning coding - Google Patents

Abstract

The present invention discloses a system for learning coding, comprising: a coding task storage unit in which n (n is a natural number) coding task code_assignments are each stored; a coding storage unit storing coding results code_nm written by m people (m is a natural number) for the nth coding task, respectively; a coding result input unit which receives the coding result code_n(m+1) written by the (m+1)th user for the nth coding task code_assignment; a code similarity determination unit which determines code similarity between the coding results of the (m+1) people; a pseudo-code set selection unit which includes code results whose code similarity is less than a threshold value in the same pseudo-code set; a coding score storage unit which stores a coding score for the coding result code_nm; and when the (m+1)th user requests it, a model answer provision unit which provides the coding result with the highest coding score among the similar code sets such as the (m+1)th user's coding result code_n(m+1). The present invention supports coding learners to understand on their own what problems exist in the code they have created.

Description

Coding learning system {SYSTEM OF LEARNING CODING}

The present invention relates to a coding learning system, and more specifically, from a learner's point of view, in order to improve my coding skills, I want to check the code of a person who wrote a code similar to myself, or a person who wrote a code with a perfect score in a different type. It relates to a coding learning system that wants to learn while looking at code, and from the standpoint of a coding teacher, can support presenting a customized model answer suitable for each of a large number of learners.

As the social awareness that coding education must be started from an early age in order to secure artificial intelligence (AI) talent has spread, software (SW) education has been made compulsory in middle schools nationwide since 2018, and has been expanded to 5th and 6th grades of elementary school from 2019. However, at the front line of education, there are no teachers to teach coding, and there is no proper curriculum, so there is a high concern that public education will be limited to 'show-off'.

In other words, learners want to improve their skills by referring to the codes of other perfect scorers, and teachers have needs in the educational field to present customized model answers suitable for a large number of learners. The reality is that there were no systems or tools.

On the other hand, in relation to the coding learning system, according to Korean Patent No. 10-1872005 (registered on June 21, 2018) (Patent Document 1), “The present invention is a method for providing a programming curriculum using a block-type programming editor, Obtaining programming behavior information of the learner; Analyzing the learner's programming behavior pattern; Corresponding to the analyzed programming behavior pattern of the learner in real time, redesigning the learner's programming curriculum; And the designed programming curriculum Discloses a method for providing a programming curriculum, including the step of providing the learner to the learner."

In addition, in relation to computer programming education, according to Korean Patent No. 10-1683022 (Announced on November 30, 2016) (Patent Document 2), "Online coding learning method, server and system are provided. Online coding learning method Receiving a first code from a first learning terminal accessing an online coding learning server; Receiving a modification request for the first code from the first learning terminal; Sending the first learning terminal to the online coding learning server. Selecting a second learning terminal for modifying the first code from among one or more learning terminals accessed at the same time; providing the first code to the selected second learning terminal; and the second learning terminal. and receiving the first code corrected by the second learning terminal as a second code from and providing the first code to the first learning terminal."

However, in Patent Literature 1 and Patent Literature 2, the learner wants to see the most similar type of perfect scorer code or refer to other types of perfect scorer codes to understand what is wrong with the code I created, However, from the standpoint of coding teachers, there seems to be no problem at all about the coding learning system, which can support presenting customized model answers for each learner.

Republic of Korea Patent No. 10-1872005 (Registered on June 21, 2018), Title of Invention: Programming Curriculum Provision Method {METHOD OF PROVIDING PROGRAMMING CURRICULUM} Republic of Korea Patent No. 10-1639301 (2016.07.07. Announcement), Title of Invention: Online coding learning method, server and its system {ONLINE CODING LEARNING METHOD, SERVER AND SYSTEM THEREOF}

The present invention was created to solve the above-mentioned problems, and the present invention is for learners who want to check the code of a person who wrote a code similar to their own in order to improve their skills, or who wrote a code with a perfect score in a different type. Looking at , I want to understand what problems are in my code, and from the perspective of a coding teacher, I want to launch a coding learning system that can support the presentation of customized model answers for each of a large number of learners. The purpose.

The coding learning system according to the present invention for achieving the above object is a coding task storage unit in which n (n is a natural number) coding task code_assignment is stored, and m (m is a natural number) for the nth coding task. A coding storage unit in which each coding result code_nm written by ) is stored; a coding result input unit for receiving a coding result code_n(m+1) written by the (m+1)th user for the nth coding task code_assignment; a code similarity determining unit for determining code similarity of the (m+1) coding results with respect to each other; a similar code set selection unit for including the code results having the code similarity smaller than a threshold value in the same similar code set; a coding score storage unit to store coding scores for the coding result code_nm; and when requested by the (m+1)th user, a best answer that provides a coding result having the highest coding score among similar code sets such as the coding result code_n(m+1) of the (m+1)th user. Including;

Preferably, the code similarity determination unit calculates a code similarity distance between the (m+1) coding results code_nm for the n-th coding task.

Here, the code similarity distance calculates the distance between the code result code_nk and the code result code_nj. A variable is removed from each of the code result code_nk and the code result code_nj, and a constant is set in each of the code result code_nk and the code result code_nj. , leaving only the structures of the code result code_nk and the code result code_nj, and calculating the number of structures having the difference by obtaining a difference between the structures of the code result code_nk and the code result code_nj, where the difference is I refers to a structure that is different from each other, or a structure that is present in one code result but not in the other code result, k is a natural number of 1≤k≤m+1, and j is 1≤j≤m+1 is the natural number of

In addition, it is preferable to automatically adjust the threshold value so that three or four pseudocode sets appear.

In addition, it is preferable to further include a coding score scoring unit for scoring a coding score for the coding result code_nm.

In addition, it further includes; a coding score scoring unit for scoring a coding score for the coding result code_nm based on the number of correct output values by inputting a certain number of input values for the coding result code_nm. desirable.

According to the coding learning system according to the present invention,

First, from a coding learner's point of view, it satisfies the need to know how to write well-written codes by other learners.

Second, it supports coding learners to understand what problems are in their code.

Specifically, showing one code of a perfect scorer belonging to a code set similar to the learner or user is more helpful in understanding whether one's code is not perfect than showing the code of a perfect scorer belonging to a different code set. It is like showing a model answer, that is, it is a code that is close to the learner's code, so it can be a lot of reference for the learner.

On the other hand, showing code written in a completely different approach other than the way or approach written by the learner or user can help learners improve their coding skills, but this is code that is close to the learner's code, that is, similar to each other. Showing the code in the code set and then showing or telling it is more effective for learning.

Third, from a coding teacher's point of view, it is possible to support presenting a customized model answer for each learner. That is, it is possible to show the perfect score code within the pseudo code set to which the learner belongs, instead of presenting the same model answer for all.

1 shows examples of three coding samples (sample 1, sample 2, and sample 3) stored in the coding storage of the coding learning system according to the present invention.
FIG. 2 shows a conversion of sample code No. 1 of FIG. 1 into an Abstract Syntax Tree (AST).
FIG. 3 shows a conversion of sample code No. 2 of FIG. 1 into an Abstract Syntax Tree (AST).
4 and 5 show conversion of sample code No. 3 of FIG. 1 into an Abstract Syntax Tree (AST).
6 illustrates a process of converting AST to Finger_print.
FIG. 7 shows conversion of the AST of sample No. 1 in FIG. 2 to Finger_print.
8 shows conversion of the AST of sample 2 in FIG. 3 to Finger_print.
9 and 10 show conversion of the AST of sample 3 in FIGS. 4 and 5 to Finger_print.
11 illustrates a process of calculating a distance between codes.
12 is a calculation of a distance between sample code 1 and sample code 2.
13 is a result of calculating distances between each of sample code No. 1, sample code No. 2, and sample code No. 3.
FIG. 14 shows that, based on the results of FIG. 13 , the conclusion that samples 1 and 2 are more similar than sample 3 is reached.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Prior to this, the terms or words used in this specification and claims should not be construed as being limited to the usual or dictionary meaning, and the inventor appropriately uses the concept of the term in order to explain his/her invention in the best way. Based on the principle that it can be defined, it should be interpreted as meaning and concept consistent with the technical spirit of the present invention.

Therefore, since the embodiments described in this specification and the configurations shown in the drawings are only one of the most preferred embodiments of the present invention and do not represent all of the technical ideas of the present invention, various alternatives may be used at the time of this application. It should be understood that there may be equivalents and variations.

Hereinafter, a coding learning system according to the present invention will be described with reference to the drawings.

A coding learning system according to the present invention includes a coding task storage unit; coding storage unit; Coding result input unit; a code similarity determination unit; a similar code set selector; coding score storage unit; and a model answer providing unit.

First, the coding task storage unit stores n (n is a natural number) coding task code_assignment. It is possible for a learner (user) to access the coding learning system and learn n coding tasks by himself.

Next, the coding storage unit stores coding result code_nm written by m people (m is a natural number) for the n-th coding task. This coding result code_nm may be stored in advance or may be a coding result additionally input while the coding learning system is operated.

Next, the (m+1)th user (learner) receives the coding result code_n(m+1) written for the nth coding task code_assignment.

The coding result code_n(m+1) entered in this way is generated by the coding learning system according to the user's or learner's request, automatically, or when a specific requirement is met, as needed by the coding learning system (m+1). It has a configuration for determining the code similarity of the coding results with respect to each other.

The pseudocode set selector is a component that includes code results having the aforementioned code similarity smaller than a threshold value in the same pseudocode set.

Also, the coding score storage unit is a component that stores coding scores for the above-described coding result code_nm.

In addition, when the (m+1)th user requests, the model answer providing unit has the highest coding score or the highest coding score among similar code sets such as code_n(m+1), the coding result written by the user. In the case of a plurality of high coding results, one or more of them is provided.

In addition, the code similarity determiner preferably calculates and determines a code similarity distance between (m+1) coding results code_nm for the n-th coding task.

Here, the code similarity distance calculates the distance between the code result code_nk and the code result code_nj. A variable is removed from each of the code result code_nk and the code result code_nj, and a constant is set in each of the code result code_nk and the code result code_nj. removed, leaving only the structure of the code result code_nk and the code result code_nj (the structure of the code result refers to an Abstract Syntax Tree (AST), which will be described later), and in the structure of the code result code_nk and the code result code_nj. The difference is obtained and the number of structures with the difference is calculated, where k is a structure that is different from each other or a structure that is present in one code result but not in the other code result. It is a natural number of 1≤k≤m+1, and j is a natural number of 1≤j≤m+1. The part described with reference to the figure below for obtaining the difference in code structure and finding the distance between codes using the structure with the difference (change and deletion operations occur by converting AST to Finger_print and comparing Finger_print respectively) Calculating the number as a distance) will be explained in detail.

In addition, it is preferable to automatically adjust the threshold value so that three or four pseudocode sets appear.

In addition, it is preferable to further include a coding score scoring unit for scoring a coding score for the coding result code_nm. In addition, it further includes; a coding score scoring unit for scoring a coding score for the coding result code_nm based on the number of correct output values by inputting a certain number of input values for the coding result code_nm. desirable.

For example, for a multiplication table code result, it is possible to enter all or part of the multiplication table as an input value, see if an accurate result is obtained, and score the perfect score based on the ratio of the correct value. Alternatively, it is also possible to score based on weights using the incorrect answer ratio for those having a high error rate in the coding result code_nm.

For example, for three input values, assuming that the error rate is 25% for input value 1, 50% for input value 2, and 75% for input value 3, (1) input value 1 For coding results that output accurate values only for values, 1 million points * {25 / (25+50+75)} = 100/6 = 16.7 points, (2) Correct values for input values 1 and 2 For the output coding result, 1 million points * {(25+50) / (25+50+75)} = 100*3/6 = 50 points, (3) Coding that outputs the correct value for input value 3 For the output, 1 million points * {(75) / (25+50+75)} = 100 * 3/6 = 50 points, (4) For the coding output that outputs accurate values for input values 2 and 3 For, 1 million points * {(50 + 75) / (25 + 50 + 75)} = 100 * 5/6 = 83.4 points, that is, it is possible to increase the weight (point distribution) for input values with a high error rate. If you match the input value with a high error rate, the score, which is the result of scoring, will be high.

Next, below, the coding learning system according to the present invention will be described with three code samples.

1 shows three coding samples (sample No. 1, sample No. 2, and sample No. 3) stored in the coding storage unit of the coding learning system according to the present invention. The three coding samples are codes written by three learners or users for the same coding task code_assignment.

FIG. 2 shows the conversion of sample code No. 1 of FIG. 1 into an Abstract Syntax Tree (AST), FIG. 3 shows the conversion of sample code No. 2 of FIG. 1 into an Abstract Syntax Tree (AST), and FIG. 4 and FIG. 5 shows conversion of sample code No. 3 of FIG. 1 into an Abstract Syntax Tree (AST).

And, FIG. 6 shows a process of converting AST to Finger_print. As shown in FIG. 6, each AST node is pre-traversed (when nodes are configured in a tree structure, the parent node is visited first and then the child node is visited in the order), the first letter of the visited node and the next letter _ extract If the number of extracted characters is less than 4, the empty space is filled with _. It will iterate over the entire AST. Attribute is extracted for each node. In the case of a node corresponding to a variable, the attribute refers to the variable type (eg, int in C language), and in the case of an operator node, the type of actual operator (eg ++, +=, x) correspond to this.

Applying this method, FIG. 7 shows the conversion of the AST of sample No. 1 in FIG. 2 into Finger_print, and FIG. 8 shows the conversion of the AST of sample No. 2 in FIG. 3 into Finger_print. 10 shows conversion of the AST of sample 3 in FIGS. 4 and 5 to Finger_print.

Next, FIG. 11 illustrates a process of calculating a distance between codes in order to determine a degree of similarity between codes. As shown in FIG. 11, Finger_print is compared for each sample in pairs. When a change or delete operation occurs, the distance increases by one. Finally, the distance between the two code samples is calculated by calculating the number of operations for the two samples to be the same.

For example, FIG. 12 is a calculation of the distance between sample code 1 and sample code 2. As shown in FIG. 12, in order to calculate the distance between the code of sample 1 and the code of sample 2, In order for the Finger_print of the code sample and the Finger_print of the second code sample to be the same, a total of 15 changes and deletions are required, so the distance between the code sample code No. 1 and the sample code No. 2 becomes 15.

13 is a result of calculating distances between each of sample code No. 1, sample code No. 2, and sample code No. 3. When determining the code similarity, all mutual distances between codes are calculated. The distance between sample code No. 1 and code No. 2 is 15, the distance between sample code No. 1 and sample code No. 3 is 35, and the distance between sample code No. 2 and sample code No. 3 is 35.

Next, FIG. 14 shows that based on the results of FIG. 13 , the conclusion that samples 1 and 2 are more similar than sample 3 is reached. As shown in FIG. 14, the distance between sample code No. 1 and sample code No. 2 is 15, and the distance between sample code No. 1 and sample code No. 3 is 35, so that code sample No. 1 and code sample No. 2 are more similar codes. can know that

In this way, based on the distance between codes, the code similarity is determined, and the similar code set is about 3 or 4 (of course, 3 or 4 may be changed according to the design of the coding learning system), the threshold By setting a value distance, codes having a distance smaller than the threshold value are selected to be included in the pseudo code set.

In the case of using the coding learning system problem according to the present invention, it is possible for a learner or user to inform which pseudo code set the code I wrote belongs to, and if the learner or user does not receive a perfect score for the coding task, No matter how hard you try, there may be cases where you only get 90 points out of 100. For example, if one conditional statement is omitted.

In this case, showing one code of a perfect scorer belonging to a code set similar to the learner or user is more helpful in understanding whether one's code is not perfect than showing the code of a perfect scorer belonging to a different code set. It is like showing a model answer, that is, it is a code that is close to the learner's code, so it can be a lot of reference for the learner.

On the other hand, showing code written in a completely different approach other than the way or approach written by the learner or user can help learners improve their coding skills, but this is code that is close to the learner's code, that is, similar to each other. Showing the code in the code set and then showing or telling it is more effective for learning.

In addition, from the standpoint of a coding teacher, it is possible to support presenting a customized model answer for each of a plurality of learners. That is, it is possible to show the perfect score code within the pseudo code set to which the learner belongs, instead of presenting the same model answer for all.

As described above, although the present invention has been described by the limited embodiments and drawings, the present invention is not limited thereto, and the technical spirit of the present invention and the following by those skilled in the art to which the present invention belongs Of course, various modifications and variations are possible within the scope of equivalents of the claims to be set forth.

Claims (5)

A coding task storage unit in which n (n is a natural number) coding task code_assignment is stored, respectively;
For the n-th coding task, a coding storage unit storing coding result code_nm written by m people (m is a natural number);
a coding result input unit for receiving a coding result code_n(m+1) written by the (m+1)th user for the nth coding task code_assignment;
a code similarity determining unit for determining code similarity of the (m+1) coding results with respect to each other;
a similar code set selection unit for including the code results having the code similarity smaller than a threshold value in the same similar code set;
a coding score storage unit to store coding scores for the coding result code_nm; and
When requested by the (m+1)th user, a best answer that provides a coding result with the highest coding score among similar code sets such as the coding result code_n(m+1) of the (m+1)th user. Including;
Coding learning system.
According to claim 1,
The code similarity determination unit,
Characterized in that for the n-th coding task, a code similarity distance is calculated for the (m + 1) coding results code_nm to each other,
Coding learning system.
According to claim 2,
The code similarity distance,
Calculate the distance between the code result code_nk and the code result code_nj,
removing a variable from each of the code result code_nk and the code result code_nj;
By removing a constant from each of the code result code_nk and the code result code_nj,
Leaving only the structure of each of the code result code_nk and the code result code_nj,
By obtaining a difference in the structure of the code result code_nk and the code result code_nj,
Calculate the number of structures with the difference,
Here, the structure with the difference refers to a different structure or a structure that exists in one code result but not in the other code result,
k is a natural number of 1 ≤ k ≤ m + 1, j is a natural number of 1 ≤ j ≤ m + 1,
Coding learning system.
According to claim 1,
Characterized in that the threshold value is automatically adjusted so that three or four pseudo code sets appear.
Coding learning system.
According to claim 1,
Further comprising: a coding score scoring unit for scoring a coding score for the coding result code_nm;
Characterized in that the coding score scoring unit scores a coding score for the coding result code_nm based on the number of correct output values by inputting a certain number of input values for the coding result code_nm,
Coding learning system.

Images