KR102550868B1 - verification system for achievements of faculty - Google Patents

Abstract

The present invention relates to a teacher achievement verification system, and more particularly, by automatically performing the entire process of verifying achievement information entered by a teacher, not only can a considerable amount of manpower, time and cost be saved, but also It is about a teacher achievement verification system that can be verified more accurately by determining similarity rather than identity.
The present invention for achieving the above object is an information input step in which teachers input achievement information, an information extraction step in which achievement-related information is extracted from the web through the input achievement information, and achievement-related information extracted in the information extraction step. It is characterized in that it consists of a data verification step of comparing and verifying achievement information.

Description

Verification system for achievements of faculty}

The present invention relates to a teacher achievement verification system, and more particularly, by automatically performing the entire process of verifying achievement information entered by a teacher, not only can a considerable amount of manpower, time and cost be saved, but also It is about a teacher achievement verification system that can be verified more accurately by determining similarity rather than identity.

In order to verify the achievements of faculty members working in universities, etc., it is necessary to search and compare the contents of thesis, patent, book, or translation written by each faculty member. More than 400 cases, and more than 200 cases in the case of translations or books.

In order to verify such a considerable amount of theses, patents, books or translations, a large number of workers manually visit each website and verify it, and it is not done at once but cross-validated, which consumes a large number of manpower. Because of this, it costs a lot of money.

Therefore, when the above-described verification work is performed automatically using a specific system instead of manually, there is an effect of saving manpower, time, and cost.

As an example of a technology for automatically performing this verification task, there is a technology described in Korean Patent Publication No. 10-2020-0082218 as shown in FIGS. 1 and 2, the technical feature of which is to increase reliability of unreliable data. extracting keywords from postings posted by individuals on SNS by the verifying server; crawling news articles containing the keyword on the web; and evaluating reliability of posts posted by the individual on SNS based on the number of crawled news articles.

However, the technology described in Korean Patent Publication No. 10-2020-0082218 is a technology that automatically verifies the reliability of the contents of posts uploaded to SNS by crawling news articles published on the web, which automatically verifies the uploaded information. Although there is an advantage, there is a problem in that it cannot accurately determine when a keyword is modified due to various reasons because only cases in which keywords exactly match are considered during verification.

Korean Patent Publication No. 10-2020-0082218 (published on July 8, 2020) Korean Registered Patent No. 10-1153138 (registered on May 29, 2012)

The present invention has been devised to solve the above problems, and an object of the present invention is to extract information from a website containing the information based on achievement information written by teachers through an extraction module having a crawler function to obtain achievement information. It is to provide a teacher achievement verification system that can significantly reduce manpower, time and cost by automatically performing the entire process by automatically comparing with

And, another object of the present invention is to compare achievement information with information extracted from each website through an extraction module, if the achievement information is alphabetic, convert it to all lowercase letters, and if it is Korean, go through a preprocessing process that decomposes into consonants and vowels It is to provide a teacher achievement verification system that can increase the accuracy of verification results by adding additional points to continuity while verifying based on sequence matching during verification.

The present invention to solve these problems;

The input achievement information DB storage step of storing the achievement information input by teachers in the teacher DB, the information extraction step of extracting achievement-related information from the web through the input achievement information, and the achievement-related information extracted in the information extraction step and the above It is characterized in that it consists of a data verification step of verifying by comparing input achievement information.

Here, the input achievement information is characterized in that each teacher inputs information about his/her thesis, patent, book, or translation.

In addition, the information extraction step includes an access step of accessing a website capable of obtaining information on papers, patents, books, or translations through an extraction module provided in the server, and crawling the information from the accessed website. It is characterized in that it consists of a data extraction step of extracting and a data refinement step of extracting only achievement-related information from the information extracted in the data extraction step.

At this time, the data trimming step is characterized by removing irrelevant characters including special characters other than achievement-related information from the extracted information by referring to the format of each website.

Meanwhile, the data verification step includes a preprocessing step of converting and processing the input achievement information stored in the teacher DB and the achievement related information extracted through the information extraction step into a set format, and the preprocessed achievement related information and the input achievement information It is characterized in that it consists of a comparison step of comparing and verifying.

Here, in the pre-processing step, if the input achievement information and achievement-related information are alphabets, they are converted to lowercase letters, and if they are Korean letters, they are separated into consonants and vowels.

At this time, the comparison step is characterized in that the achievement information and achievement-related information are compared by adding a weight to continuity in a Levenstein Distance algorithm among sequence matching methods.

In addition, the comparison step is characterized in that the similarity is determined by the weight edit distance value (D(i,j)) derived through the following process.

[procedure]

1. If A[i] and B[j] of the string to be compared match,

D(i,j) = D(i-1,j-1),

If b == True then w = w + 1,

If b != True then b = true.

2. If A[i] and B[j] of the string to be compared do not match,

D(i,j) = min( D(i-1,j)+1/w, D(i,j-1)+1/w, D(i-1,j-1)+1/w ) ego,

b = False.

{ D(i,j) = weight edit distance value, initial value D(0,0)=0,

A[i] = ith character of string A,

B[j] = jth character of string B,

w = continuity weight (initial value of w is 0),

b = Boolean to determine continuity (initial value of b is False),

i,j = 0 to n }

According to the present invention having the above configuration, based on the achievement information written by the teachers, information is extracted through an extraction module having a crawler function from a website having the corresponding information and automatically compared with the achievement information, thereby completing the entire process Since it can be performed automatically, it has the effect of significantly saving manpower, time and cost.

In the present invention, when comparing information extracted from each website and achievement information through an extraction module, if the achievement information is in the alphabet, all of it is converted to lowercase letters, and if it is in Korean, it is subjected to a preprocessing process of decomposing into consonants and vowels, At the time of verification, a string comparison method (sequence matching) is used as a basis, but additional points are added to continuity to verify according to similarity, thereby increasing the accuracy of verification results.

1 is a schematic diagram of a conventional verification method through crawling.
2 is an exemplary view of a web page for extracting data including keywords in a conventional verification method through crawling.
3 is a conceptual diagram of a teacher achievement verification system according to the present invention.
4 is a block diagram of a teacher achievement verification system according to the present invention.
5 is a flowchart of a teacher achievement verification system according to the present invention.
6 is an exemplary diagram of a general Levenstein Distance, which is an example of a sequence matching technique.
7 is an exemplary view of a screen showing results upon completion of verification in the teacher achievement verification system according to the present invention.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. The same reference numerals are used for the same components in the drawings, and redundant descriptions of the same components are omitted. And, it should be understood that the present invention may be embodied in many different forms and is not limited to the described embodiments.

Figure 3 is a conceptual diagram of the teacher achievement verification system according to the present invention, Figure 4 is a block diagram of the teacher achievement verification system according to the present invention, Figure 5 is a flowchart of the teacher achievement verification system according to the present invention, Figure 6 is a sequence This is an example of a general Levenstein Distance, which is an example of a matching technique, and FIG. 7 is an example of a screen showing results upon completion of verification in the teacher achievement verification system according to the present invention.

The present invention relates to a teacher achievement verification system, and as shown in Figs. It consists of an information extraction step (S200) of extracting achievement-related information from the web through the input achievement information and a data verification step (S300) of verifying by comparing the achievement-related information extracted in the information extraction step (S200) with the input achievement information. .

Here, as shown in FIGS. 3 and 4, teachers access the server 100 equipped with the communication module 120 through a terminal 200 such as a smart phone, a desktop computer, or a tablet that each teacher possesses. Thus, each teacher's achievement information is entered.

At this time, the server 100 is provided with a teacher DB 110 to store achievement information input by each teacher by accessing the server 100. Such teacher input achievement information is usually a thesis, patent, book or Information about the epigraph is included.

In addition, the information extraction step (S200) includes the access step (S210) of accessing the website 300 from which information on thesis, patent, book or translation can be obtained, and the crawling operation on the accessed website 300. It consists of a data extraction step (S220) of extracting information through a data extraction step (S220) and a data trimming step (S230) of extracting only achievement-related information from the information extracted in the data extraction step (S220).

Here, the server 100 is provided with an extraction module 130. The extraction module 130 performs a crawler function to extract information from the website 300 through a crawling operation. The extraction module ( 130) is provided with a web DB 134 that stores information on the web site 300 where information on papers, patents, books or translations can be obtained.

So, in the connection step (S210), the extraction module 130 accesses the corresponding website 300 through the website information stored in the web DB 134, and in the data extraction step (S220), as described above Likewise, information related to the input achievement information is extracted through the crawling operation and stored in the temporary storage unit 132 provided in the extraction module 130 .

On the other hand, in the data trimming step (S230), a trimming operation is performed by deleting information other than the achievement-related information of the teachers. Deletes missing characters.

That is, there is a difference in the format of displaying each field on each website 300. For example, even when displaying a title in a thesis, a colon (:) or semicolon (;) may be used before and after the title, You can also use quotation marks (") or other special characters, or multiple spaces between each special character and the title.

Here, in the data trimming step (S230), format information of each web site 300 is included in the web DB 134, and special characters other than achievement-related information are extracted from the information extracted by the extraction module 130 by referring to this format information. This will remove extraneous characters including .

The achievement-related information processed by the extraction module 130 is stored in the temporary storage unit 132 provided in the extraction module 130 and is more accurately compared with the input achievement information in the data verification step (S300) to be described later. be able to compare.

That is, in the present invention, when comparing the achievement-related information extracted from the website 300 and the achievement information input by the teacher, a sequence matching method is used. Since it is determined whether they match by comparing them as they are, if the string to be compared contains special characters constituting the format of the web site 300, it is difficult to accurately compare them, so these meaningless characters are removed to increase accuracy in comparison.

In addition, the data verification step (S300) is a preprocessing step (S310) of changing the input achievement information stored in the teacher DB 110 and the achievement-related information extracted through the information extraction step (S200) into a set format and processing it. It consists of a comparison step (S320) of verifying by comparing the preprocessed achievement-related information with the input achievement information.

Here, in the preprocessing step (S310), if the input achievement information stored in the teacher DB 110 and the achievement-related information extracted through the extraction module 130 are alphabets, they are all converted to lowercase letters and unified, and if they are Korean, All letters are separated into consonants and vowels and stored in a separately allocated part of the teacher DB 110.

In other words, in the case of Hangeul, for example, when comparing Hong Gil- dong and Hong Gal- dong, the middle one character is different among the three letters, so it looks quite different because 1/3 is different, but when it is broken down into consonant and vowel units, Since ㅇㄱㅣㄹㄴㅗㅇ and ㅎㅗㅇㄱ아ㄹㅇ are compared, only one phoneme is different among 9 phonemes, so the similarity appears high because 1/9 is different.

In addition, in the case of alphabets, initials may be written as initials, so conversion to initials may be stored in an additionally allocated portion of the teacher DB 110.

Meanwhile, in the comparison step (S330), a weight is added to continuity in a Levenstein Distance algorithm among a sequence matching method, and the input achievement information and achievement-related information are compared and verified.

Here, the general edit distance (Levenstein Distance) algorithm compares two strings to determine how many times one of the three operations of insertion, deletion, and replacement must be performed to create the same string. will be.

Here's a programmatic representation of this process:

1. If A[i] and B[j] of the string to be compared match,

D(i,j) = D(i-1,j-1),

2. If A[i] and B[j] of the string to be compared do not match,

D(i,j) = min( D(i-1,j), D(i,j-1), D(i-1,j-1) ).

At this time, D(i,j) = edit distance value (how many times the above 3 operations are performed), the initial value D(0,0)=0, and A[i] in two character strings A,B = i-th character of character string A, B[j] = means the j-th character of character string B, and the variables i and j are integers in the range of 0 to n, and are the number of characters constituting each character string.

So, in order to compare each character constituting the string, the above process of solving the Levenstein Distance algorithm can be performed to derive the edit distance value (D(i,j)), and the edit distance value is as described above. It indicates how many times one of the three operations described above should be performed in order to compare two character strings and change them to be the same. The larger the editing distance value, the lower the similarity between the two character strings.

An example of such an edit distance (Levenstein Distance) algorithm can be clearly seen by working with a table comparing the string ABC and the string QWC as shown in FIG.

However, in the general Levenstein Distance algorithm, only a few characters with differences can be identified, and a character string with a high similarity cannot be identified even though the number of characters with differences is large.

For example, co m pa r is o n and co n pa l is e n are completely unrelated strings, but there is a difference in the underlined 3 parts, so the edit distance is 3, and compar ison and compar e do not have much similarity. The two strings of dumplings have a difference in the underlined 4 parts, so the editing distance becomes 4, so it is judged that the similarity is low.

Therefore, in the present invention, although the Levenstein Distance algorithm is used to solve this problem, the similarity is determined by adding a weight to the continuity as described above considering that the similarity is higher as the number of continuously matched parts increases. will do

And, looking at the programmatic representation of the process of adding weights to the Levenstein Distance algorithm:

1. If A[i] and B[j] of the string to be compared match,

D(i,j) = D(i-1,j-1),

If b == True then w = w + 1,

If b != True then b = True.

2. If A[i] and B[j] of the string to be compared do not match,

D(i,j) = min( D(i-1,j)+1/w, D(i,j-1)+1/w, D(i-1,j-1)+1/w ) ego,

b = False.

{ D(i,j) = weight edit distance value, initial value D(0,0)=0,

A[i] = ith character of string A,

B[j] = jth character of string B,

w = continuity weight (initial value of w is 0),

b = Boolean to determine continuity (initial value of b is False),

i,j = 0 to n }

Here, comparing two strings with the weighted edit distance (Levenstein Distance) algorithm of the present invention,

1. co m pa r is o n VS co n pa l is e n (requires 3 operations)

Edit distance value = 1 + 1/2 + 1/3 = 11/6, which is about 1.83.

At this time, look at the derivation process of the value: The value of w increases when there are consecutively matched characters (if the value of b at the time of comparison is True), and the initial value of the edit distance value is 0.

When comparing the character 'c', since the two characters are identical, D(1,1)=D(0,0)=0, and since the value of b is the initial value of False, the value of w is 0, and the value of b is then changed to true.

When comparing the character 'o', since the two characters are identical, D(2,2)=D(1,1)=0, and since the value of b is True, the value of w is incremented by 1 to become 1, and the value of b becomes 1. It doesn't change.

When comparing the character 'm', the two characters do not match, so D(3,3) = D(2,2)+1/w = 0+1/1, the value of b is changed to False, and the value of w is unchanged and is still 1.

When comparing the character 'p', since the two characters are identical, D(4,4)=D(3,3)=1, and since the original value of b is False, the value of w remains unchanged and is 1, and b is changed to true. .

When comparing the character 'a', since the two characters are identical, D(5,5)=D(4,4)=1, and since the original value of b is True, the value of w is increased by 1 to become 2, and b is It doesn't change.

When comparing the character 'r', the two characters do not match, so D(6,6) = D(5,5)+1/w = 1+1/2, the value of b is changed to False, and the value of w is unchanged and is still 2.

When comparing the character 'i', since the two characters are identical, D(7,7) = D(6,6) = 1+1/2, and since the original value of b is False, the value of w remains unchanged and is 1, and b is change to true

When comparing the character 's', since the two characters are identical, D(8,8)=D(7,7)=1+1/2, and since the original value of b is True, the value of w is increased by 1 to become 3. and b does not change.

When comparing the character 'o', the two characters do not match, so D(9,9) = D(8,8)+1/w = 1+1/2+1/3, and the value of b changes to False , and the value of w does not change, so it is still 3.

When comparing the character 'n', since the two characters are identical, D(10,10) = D(9,9) = 1+1/2+1/3, and since the original value of b is False, the value of w does not change. 3 and b is changed to true.

Therefore, the D(10,10) value, which is the edited distance value to which the final weight is applied, becomes 1+1/2+1/3.

2. compar ison VS compar e (requires 4 operations)

Edit distance value = 1/5 + 1/5 + 1/5 + 1/5 = 4/5, which is 0.8.

Therefore, when the two strings are compared with the Levenstein Distance algorithm to which the weight is added according to the present invention, the second case has a lower edit distance value of 0.8, which is judged to have high similarity, which is consistent with actual similarity. It can be seen that the result of

Therefore, in the present invention, in the comparison step (S330) constituting the data verification step (S300), the achievement information is compared with the achievement-related information using the Levenstein Distance algorithm to which a weight is added, so that when the teacher inputs or each Even if there is a typo in writing on the website, the accuracy of verification can be increased through similarity.

And, after the data verification step (S300), a verification result checking step (S400) for confirming the verification result may be further performed.

Here, in the data verification step (S300), information on the part that does not match during verification is stored in the result storage unit 142 provided separately in the verification module 140, and in the verification result confirmation step (S400) Referring to the information stored in the result storage unit 142, the result is output with a brief comment as shown in FIG.

At this time, in the verification result confirmation step (S400), if there is an error in the input data, the result information is notified to the teacher through the information stored in the corresponding teacher DB 110 so that the error can be corrected.

Although preferred embodiments of the present invention have been described above, the scope of the present invention is not limited thereto, and the spirit of the present invention extends to those within the scope substantially equivalent to the embodiments of the present invention. Various modifications and implementations are possible by those skilled in the art to which the present invention pertains within the scope that does not deviate from the above.

The present invention relates to a teacher achievement verification system, and more particularly, by automatically performing the entire process of verifying achievement information entered by a teacher, not only can a considerable amount of manpower, time and cost be saved, but also It is about a teacher achievement verification system that can be verified more accurately by determining similarity rather than identity.

100: Server 110: Faculty DB
120,210: communication module 130: extraction module
140: verification module 200: terminal
220: display unit 300: website
S100: information input step
S200: information extraction step
S300: Data Verification Step
S400: verification result confirmation step

Claims (8)

An input achievement information DB storage step in which achievement information input by teachers is stored in a teacher DB provided in the server; an information extraction step in which an extraction module provided in the server extracts achievement-related information from the web through the input achievement information; , a data verification step in which the verification module provided in the server compares the achievement-related information extracted in the information extraction step with the input achievement information to verify, and a verification result confirmation step in which verification results are confirmed,
The input achievement information is information about their thesis, patent, book, or translation entered by each teacher,
The information extraction step includes an access step in which the extraction module accesses a website where information on thesis, patent, book, or translation can be obtained, and information is extracted from the website accessed by the extraction module through a crawling operation. a data extraction step, and a data refinement step for the extraction module to extract only achievement-related information from the information extracted in the data extraction step;
The data verification step includes a preprocessing step in which the verification module converts the input achievement information stored in the teacher DB and the achievement-related information extracted through the information extraction step into a set format and processes it, and the verification module relates to the preprocessed achievement It consists of a comparison step of comparing and verifying information and input achievement information,
In the data verification step, information on a part in which the preprocessed achievement-related information and the input achievement information do not match is stored in a separately provided result storage unit;
In the verification result confirmation step, if there is an error in the achievement information input by the teachers, the result information stored in the result storage unit is output with a brief comment to inform the teacher so that the teacher can correct the error. and
In the comparison step, when the verification module compares the input achievement information and the achievement-related information in the Levenstein Distance algorithm among the sequence matching methods, a weight is added to continuity, which is the number of consecutively matched parts, Compare the achievement information and achievement-related information,
The comparison step is characterized in that the similarity is determined by the weight edit distance value (D(i,j)) derived through the following process.
[procedure]
1. If A[i] and B[j] of the string to be compared match,
D(i,j) = D(i-1,j-1),
If b == True then w = w + 1,
If b != True then b = true.

2. If A[i] and B[j] of the string to be compared do not match,
D(i,j) = min( D(i-1,j)+1/w, D(i,j-1)+1/w, D(i-1,j-1)+1/w ) ego,
b = False.

{ D(i,j) = weight edit distance value, initial value D(0,0)=0,
A[i] = ith character of string A,
B[j] = jth character of string B,
w = continuity weight (initial value of w is 0),
b = Boolean to determine continuity (initial value of b is False),
i,j = 0 to n }
delete delete According to claim 1,
In the data trimming step, the extracting module removes irrelevant characters including special characters other than achievement-related information from the extracted information by referring to the format of each website.
delete According to claim 1,
In the pre-processing step, the verification module converts the input achievement information and achievement-related information into lowercase letters if they are alphabetic, and separates them into consonants and vowels if they are Korean.
delete delete

Images