TW201316186A - Chinese anti-piracy and plagiarism detecting system and its method - Google Patents

Abstract

Here are a Chinese anti-piracy and plagiarism detecting system and its method. The system includes several mechanisms: the article loading mechanism, the article dividing mechanism, the searching and comparing mechanism, and the evaluating and reporting mechanism. All these mechanisms work via the composition of one central server, a cluster of operating host computers, and a specific search engine. While operating this system, an user first need to upload a target article which is to be detected to the central server via the article loading mechanism. The central server will distribute that target article to an operating host computer. Following, with the help of the article dividing mechanism and its own words/sentences-dividing calculating method set by the system's program, the operating host computer will divide the target article into sentence units according to the user's installation in the beginning. Next, the searching and comparing mechanism will compare, evaluate, and search out the similar web pages or articles. The identified web pages or articles found out will be downloaded entirely to the operating host computer, where the full text of the target article will be compared with the full text of identified web pages or articles. And next, the evaluating and reporting mechanism will tell and show the result of the comparison-- the identified sentences of the target article together with the identified parts of the web pages or articles will be marked simultaneously, and the source of those web pages or articles will be displayed, too. On the last step, the operating host computer will transmit the result to the central server. Therefore the user can read the result of the work.

Description

Chinese digital anti-plagiarism detection comparison system and method

The invention relates to a Chinese digital anti-plagiarism detection comparison system and method, in particular to a function of using a search engine, which is to be compared with an article to disassemble a clause after disassembling a sentence, and then search for the search result. When the summary matches the search sentence, the search results page is loaded for full-text comparison, and the Chinese digital anti-plagiarism detection comparison system and method copied from the website can be quickly determined.

In recent years, the development of the Internet has changed the habit of students to write homework. From the past, students have been encouraged to access more information on the Internet. Until now, it is necessary to prevent students from using Internet data. The use of the Internet has indeed caused a lot of plagiarism. The Internet creates a high degree of plagiarism because of the situation, so that under the pressure of situation and time, it is easy to produce plagiarism; secondly, the society permeates the psychology of "taking shortcuts", students only care about the efficiency of writing homework and not pay attention to the quality of work, the academic world also Paying attention to the publication volume of the paper, but paying less attention to the quality of the content and the mentality of the society. In addition, the biased behavior of plagiarism has become the norm. The writing service has worsened the entire academic environment. The plagiarism of dishonesty has deteriorated into academic deception. Serious crimes.

In view of the increasing plagiarism of students, especially the plagiarism of plagiarizing online data or reprocessing the network data, re-collection and arrangement of hodgepodge articles, foreign profit-making enterprises have developed digital plagiarism detection software for prevention. After a long period of operation and testing, it can really reduce the incidence of plagiarism and inhibit students' desire to take the opportunity to speculate. However, the test reports and related literatures of most detection systems are mainly based on the English-speaking countries, and all kinds of accumulated knowledge. It only stops in the English environment and cannot be transplanted to the Chinese language. Because the Chinese language is different from the English environment, regardless of the book structure, word combination, sentence sentence, and punctuation, the Chinese culture interface is very different. Digital plagiarism comparison systems and methods still need to be developed to suit the Chinese language education community.

According to the conventional Chinese plagiarism article comparison system and method, such as Taiwan Invention Patent Notice No. I262402 "Feature Extraction, Data Decryption Method, and System and Method for Plagiarized Article Search", features the article 10 that has been implanted with watermark Extracting to obtain the file feature 20, inputting the search engine 30 according to the obtained vocabulary, and comparing the index database 40 to search for articles 50 that may be plagiarized on the Internet; and then comparing the searched articles 50 with the original text, according to The watermark analysis is performed on the sentence obtained by the comparison result; finally, the obtained watermark information 60 is compared with the original watermark, and then the comparison result is used to determine whether the searched article is a plagiarized article, if the comparison result is greater than A threshold value indicates that it is a plagiarism article 70.

The feature extraction method of the conventional invention is to use the sentence and vocabulary obtained from the watermarked article to perform word segmentation and part-of-speech tagging, and then use the synonym database and synonym according to the words and sentence patterns implanted in the watermark in the article. The sentence pattern library generates the semantic level features of the article's internal text, that is, the sentence and the vocabulary with the watermark in the content are taken out. Then use the vocabulary and part of speech as the basis for the definition of the query, and search in the synonym database to obtain the vocabulary that can be used as the feature of the article. Then use the vocabulary as a keyword to use the search engine to conduct a web search to obtain relevant plagiarism articles.

However, the conventional plagiarism article search system and method have to perform the feature of extracting the article into the watermark, extracting the watermarked article, and inputting the search engine 30 and the comparison index database 40 according to the obtained vocabulary. Searching for articles 50 that may be copied on the Internet, comparing the searched articles 50 with the original text, performing a watermark analysis on the sentences obtained from the comparison results, and comparing the obtained watermark information 60 with the original watermark, And judging whether the article obtained by the search is a complicated step such as copying the article according to the comparison result, although the part of changing the synonym and the synonym sentence can be compared, but for comparing the general paper or the long article, such a large number of complicated steps are The computer server will cause a large load, which will affect the speed of the article comparison and reduce the efficiency of use.

Secondly, because the same part of the comparison system may be a "reference bibliography" or a "direct reference" that is appropriate for the scope of the comparison, it is difficult to directly determine whether an article is a plagiarism article; and finally, " The term "plagiarism" has not been defined objectively. The number of similar words is also an important reference factor for judging plagiarism. The setting of the watermarking scope does not allow users to make flexible adjustments based on their subjective definition or objective academic environment.

Therefore, in order to judge whether an article is a plagiarism article more efficiently, it is first necessary to speed up the comparison of the articles, and use the comparison result of the plagiarism comparison system as an auxiliary to judge whether it is the basis for plagiarizing the article, and finally to manually check the article. The method can be used to further determine whether an article is a copy of the article. Secondly, the user can flexibly adjust the definition of the number of words in "plagiarism" through the academic environment or needs of the user, and meet the needs of the user. Therefore, how to speed up the comparison of the articles, and let the reviewer know clearly and quickly the difference between the suspected plagiarized part and the search part in the comparison article, and the plagiarism of the plagiarized word definition as the elastic adjustment The lack of consideration in the search system.

To this end, the inventors based on years of experience in related system development and method research, specifically for the current Chinese digital anti-plagiarism detection comparison system and method to study, is the invention of the case.

The purpose of the present invention is to provide a Chinese digital anti-plagiarism detection comparison method, the user can upload the comparison article to the central server for data access, and compare and disassemble the article through the system program. The rules are disassembled, and the disassembled sentence of the fixed number of words is obtained, and then the search engine is used to search the disassembled articles one by one, and then multiple pages or articles with the same disassembled sentences can be obtained, and the same webpage or article will be used. Compared with the comparison article, the comparison result can be obtained by marking and analyzing the similar parts of the article or the similar article. The comparison result can be manually checked by the reviewer and judged. Whether the article is a copy of the article.

The Chinese digital anti-plagiarism detection comparison system of the present invention, wherein the article access mechanism sets a front-end component to provide a user to enter a Chinese digital anti-plagiarism detection comparison system pipeline, and has an upload component and a database component; When the solution mechanism starts to detect and upload the comparison article, the comparison article is decomposed into a sentence algorithm, and each comparison article is first broken into multiple paragraphs by a line feed method, and then the number of words is disassembled according to the custom The fixed number of words disassembles each paragraph into a minimum detection sentence of fixed length, and deletes the punctuation marks of each minimum detection sentence, and then compares the number of words according to the custom search, and uses the fixed number of words as the threshold to match Search for the minimum detection sentence of the comparison condition, and finally the minimum detection sentence registration number corresponding to the search comparison condition can be used to provide the search comparison mechanism for detecting plagiarism; the search comparison mechanism sets the comparison component The determination component and the plagiarism source comparison component can use the comparison component and the determination component to obtain a webpage or article that is identical to the disassembled article, and link and download with the plagiarism source comparison component. For similar webpages or articles with similar articles, the comparison article and the similar webpage or article can be compared by the comparison component and the judgment component. The evaluation report mechanism is to compare and judge the above-mentioned comparison article. , you can analyze each of the uploaded comparison articles and get the comparison results.

The Chinese digital anti-plagiarism detection comparison system and method of the present invention is provided with a plurality of computing hosts in a central server of the Chinese digital anti-plagiarism detection comparison system, and the user is uploaded by the central server to compare the articles. Distribute to each computing host, use the computing host to disassemble the article, download multiple pages and articles that are identical to the article dismantling, and compare the full text to obtain the comparison result, and return the comparison result to the central server. And then transmitted to the user via the central server.

Thus, in order for the examiner to fully understand the features of the present invention, the following is illustrated by the accompanying drawings: FIG. 1 is a flow chart showing the steps of the digital plagiarism comparison method of the present invention. First, the user can upload the comparison article 10 to the central server 11 for data access, and disassemble the comparison article 10 by disassembling the sentence algorithm to obtain the disassembled article 12, which is to be disassembled by the search engine 13. After the minimum detection sentence of the article 12 is uploaded and searched, you can obtain multiple pages or articles 14 that are identical to the disassembled article 12, and download the same page or article 14 and compare the article with The same webpage or article 14 is compared in full text 15 to compare and compare the same part of the article 10 with the same webpage or article 14, and the comparison result 16 can be obtained, and the comparison result 16 can be performed by the examiner. Manual inspection to further determine whether the comparison article is a plagiarism article.

Furthermore, the Chinese digital anti-plagiarism detection comparison system of the present invention at least includes: an article access mechanism 2, and a homepage component 20 for providing a user to enter a Chinese digital anti-plagiarism detection comparison system pipeline, including a website homepage, use The identification of the account number and password, the mechanism for forgetting the user account password, etc.; and the user identity can be divided into system administrators, teachers, students and visitors to facilitate user management and set usage rights.

The foregoing article access mechanism 2 is provided with an uploading component 21 for providing a website page for the user to upload the matching article 10 to the central server 11, and the uploading component 21 will determine whether the uploading of the article 10 format (Word or PDF) conforms. The system requirements; the upload component 21 will store the upload comparison article 10 into the database component 22 of the central server 11, and monitor the user to re-upload the comparison article 10 within the deadline.

The foregoing article access mechanism 2 is provided with a database component 22, which mainly collects and records various types of digital data on the Internet at any time to increase the source of the data during the comparison.

The article dismantling mechanism 3, the above-mentioned article dismantling mechanism 3, when starting to detect the uploading comparison article 10, the matching article 10 is to disassemble the sentence algorithm 30, and each of the matching articles 10 is firstly broken in the line-entry mode 31. In the plural paragraph 310, according to the custom disassembled word number 32, each paragraph 310 is disassembled into a fixed length minimum detection sentence 320 by a fixed number of words, and the punctuation marks of each minimum detection sentence 320 are deleted. According to the custom search comparison word number 33, the minimum number of detection sentences 320 matching the search comparison condition is selected by using the fixed word number as a threshold, and the minimum number of words remaining in the minimum detection sentence 320 after deleting the punctuation mark is small. When the searched comparison number of words is set to 33, the search sentence is not searched for the sentence; thus, the disassembled sentence algorithm 30 can be used to solve the problem that the sentence after the disassembled article 10 is too short, has no search meaning, or is compared. The sentence after the disassembly of the article 10 is too long to search for the problem; finally, the foregoing minimum detection sentence 320 corresponding to the search comparison condition can be registered to provide the search comparison mechanism 4 for detecting plagiarism.

Regarding the aforementioned disassembling sentence algorithm 30, for example, a paragraph 310 in which a certain collating article 10 is broken in a line-feeding manner 31 is: "He slowly squats down, a little better, a little better. From a slight elevation angle, he saw a strange woman holding a balloon across the street looking up at the sky. She found the sullen, and the mouth opened involuntarily. If the number of custom disassembled words 32 is 15 words, the paragraph 310 can be disassembled into a fixed length minimum detection sentence 320 as: "(He slowly squats down, a little better, one better) ) (point. From a slight elevation angle, he saw the street across the street) (a strange woman with a balloon in hand is looking up) (looking at the sky, she looks like a sulky, mouth does not) (opened by autonomously.) After that, the minimum detection sentence 320 is deleted from the punctuation mark, and the minimum detection sentence 320 without punctuation is obtained: "(He slowly squats down a little better) (point from a slight elevation angle he Seeing across the street) (a strange woman with a balloon in hand is looking up) (Looking at the sky, she looks like a sullen mouth) (by self-opening)"; if the custom search is 33 8 words, in which the sentence "(opened by autonomously)" has only 7 words, which does not meet the screening threshold. Therefore, the minimum detection sentence 320 of the last login number and upload search comparison is: "(He slowly squats down) Ok, it’s a little better.) (The point is from a slightly elevated angle. He sees the street across the street.) (The strange woman with a balloon in hand is looking up) (Wangtian She is empty and looks like a sulky mouth.) A total of four sentences.

The search comparison mechanism 4 is provided with a matching component 40. The comparison component 40 can pass through an internet search engine 13 (such as a Google search engine or a YAHOO search engine), or various electronic article databases, or articles of other students. The selected minimum detected sentence 320 is compared to the above-mentioned screened, or the matching article 10 is compared with the identical web page or the article 14 in full text 15 .

The search matching mechanism 4 is provided with a determining component 41. The function of the determining component 41 is to determine whether to plagiarize each sentence and each paragraph in the article 10 according to the comparison result of the comparing component 40. The percentage is used to present the possibility of plagiarism; or the proportion of plagiarism that compares article 10 with the same page or article 14 is also evaluated. The aforementioned search comparison mechanism 4 sets the plagiarism source comparison component 42, mainly to establish a link between the plagiarism text and the plagiarism source, to link and download the similar page or article 14 similar to the comparison article 10 for full-text comparison 15 .

The method for obtaining a web page or article 14 similar to that of the article 10 of the present invention is to use the search engine 13, the matching component 40 and the decision component 41 of the search matching mechanism 4 to filter the disassembled sentence algorithm 30. The minimum detection sentence 320 is searched and compared separately, and the search comparison result obtained after the search is calculated according to the sum of the search sentence and the search comparison result, and the ratio of the common sub-sequence is calculated. And setting a threshold value, if the proportion of the common subsequence exceeds the threshold value, that is, the plagiarism source comparison component 42 is linked and downloads the webpage or article, and the webpage or article 14 which is identical to the disassembling article 12 is obtained. For the subsequent full-text comparison 15 program.

The method of the full text comparison 15 of the present invention utilizes the matching component 40 using the search matching mechanism 4, and the comparison article 10 and the identical web page or article 14 are directly connected by meaningless characters, using the suffix array ( Suffix Array) data structure has been processed, and then using Data Partitioning (DP) techniques, you can get all the sentences that are the longest in the comparison article 10 and appear on the same page or article 14, and then The determination component 41 of the search comparison mechanism 4 is used to evaluate and determine the proportion of plagiarism after the full text comparison; since the same part of the comparison may be a "reference bibliography" or a conforming "direct citation", for the convenience of reviewers For manual inspection, the comparison of the article 10 with the same network 14 is the same portion, and the search comparison mechanism 4 is separately marked in the anti-yellow manner in the comparison article 10 and the similar web page or article 14.

The evaluation reporting mechanism 5 is the last item of all detection work, that is, after comparing and judging the above-mentioned comparison article 10 with the same webpage or article 14, the evaluation report mechanism 5 can be used for each upload. The comparison of article 10 is analyzed and the results of the comparison 16 are obtained.

As shown in FIG. 2, it is an embodiment of the Chinese digital anti-plagiarism detection and comparison system of the present invention. Each user 6 uploads the comparison article 10 to the central server 11 via the article access mechanism 2 for data access. Then, the central server 11 distributes the comparison article 10 uploaded by each user 6 to each computing host 110, and the computing host 110 uses the article disassembly mechanism 3 to disassemble the article 10 to obtain the disassembled article 12. After searching and comparing the disassembled articles 12 with the search engine 13 by using the search engine 13, the webpage or article 14 similar to the disassembled article 12 can be downloaded and sent back to the computing host 110. Compare the article 10 with the similar web page or the full text of the article 14 and use the evaluation report mechanism 5 to analyze the uploaded article 10 for each upload, and after generating the comparison result 16, A comparison result 16 is passed back to the central server 11, so that the user 6 can view the comparison result 16.

In this way, the present invention utilizes the division mechanism of the central server 11 and the computing host 110 to improve the search comparison speed of the Chinese digital anti-plagiarism detection comparison system, and the Chinese digital anti-plagiarism detection comparison system utilizes the article disassembly. Mechanism 3 and the search comparison mechanism 4 can speed up the comparison of the article 10, and the same part of the comparison article 10 and the similar network 14 are respectively marked in the anti-yellow manner in the article and the webpage, so that the user 6 can quickly obtain the comparison result 16 of the comparison article 10, and use the comparison result 16 for manual inspection to determine whether the comparison article is a plagiarism article, which is the composition of the case.

10. . . Comparison article

11. . . Central server

12. . . Disassemble the article

13. . . Search engine

14. . . Similar webpage or article

15. . . Full text comparison

16. . . Comparison result

2. . . Article access mechanism

20. . . Home component

twenty one. . . Upload component

twenty two. . . Database component

3. . . Article dismantling mechanism

30. . . Disassemble word algorithm

310. . . paragraph

32. . . Disassemble words

320. . . Minimum detection sentence

33. . . Search for the number of words

4. . . Search comparison mechanism

40. . . Alignment component

41. . . Decision component

42. . . Plagiarism source comparison component

5. . . Evaluation reporting mechanism

6. . . user

110. . . Computing host

1 is a flow chart showing the steps of the Chinese digital anti-plagiarism detection comparison method of the present invention.

Figure 2 is an embodiment of the invention.

10. . . Comparison article

11. . . Central server

12. . . Disassemble the article

13. . . Search engine

14. . . Similar webpage or article

15. . . Full text comparison

16. . . Comparison result

Claims (8)

A Chinese digital anti-plagiarism detection comparison method includes the following steps: uploading a comparison article to a central server for data access; and disassembling the article by disassembling the word-breaking algorithm to obtain a disassembled article; After the search engine uploads the minimum detection sentence of the disassembled article one by one, it can obtain multiple pages or articles that are similar to the disassembled article, and download the same webpage or article; the comparison article will be identical. The webpage or the article is compared in full text, and the comparison result is marked and analyzed by comparing the similarity of the article with the similar webpage or the article, and the comparison result is obtained for the examiner to manually check and compare the article. Whether it is a copy of the article. A Chinese digital anti-plagiarism detection comparison system includes at least an article access mechanism, and a homepage component is provided to provide a user to enter a Chinese digital anti-plagiarism detection comparison system pipeline; the foregoing article access mechanism sets an upload component to provide a website The page is for the user to upload the comparison article to the central server; the foregoing article access mechanism sets the database component to collect various types of digital data on the Internet at any time, and stores it, thereby increasing the source of the data when comparing; The solution mechanism, the above-mentioned article dismantling mechanism starts to detect and upload the comparison article, and the comparison article is disassembled into a sentence algorithm, and each comparison article is first broken into multiple paragraphs by line feed, and then customized. Disassemble the number of words, disassemble each paragraph into a fixed-length minimum detection sentence with a fixed number of words, and delete the punctuation marks of each minimum detection sentence, and then compare the number of words according to the custom search to fix the word The number is used as a threshold to filter the minimum detection sentence that meets the search comparison condition, and finally the minimum detection sentence registration number corresponding to the search comparison condition can be selected to provide a search. The comparison mechanism detects the plagiarism; the search comparison mechanism sets the comparison component, and the comparison component can be logged in through the Internet search engine, or various electronic article databases, or other student articles. The minimum detected sentence is selected, or the comparison article is compared with the identical webpage or article; the search matching mechanism is provided with a determining component, and the function of the determining component is based on the comparison result of the comparing component. Evaluate whether or not to plagiarize each sentence and each paragraph in the article, and present the possibility of plagiarism by percentage, or to compare the plagiarism ratio of the comparison article with the same page or article. Judgment; the above-mentioned search comparison mechanism sets the plagiarism source comparison component, mainly to establish a link between plagiarism text and plagiarism source, to link and download the same page or article similar to the comparison article for full-text comparison; evaluation report mechanism Is the last item of all detection work. After comparing and judging the above-mentioned comparison articles, you can target each one. Than to make the article analyzed and compared with the results obtained. According to the Chinese digital anti-plagiarism detection and comparison system described in item 2 of the patent application scope, the homepage components include the homepage of the website, the identification of the user account and password, and the mechanism for forgetting the user account password, etc.; The identity of the person is divided into system administrators, teachers, students and visitors to facilitate user management and set usage rights. According to the Chinese digital anti-plagiarism detection and comparison system described in item 2 of the patent application scope, the uploading component will determine whether the format of the uploading comparison article conforms to the system requirements, and the uploading component will merge the uploading comparison article storage into the center. The server's database component, and the monitoring user can re-upload the comparison article within the deadline. According to the Chinese digital anti-plagiarism detection and comparison system described in the first or second patent application scope, the method for obtaining the same web page or article similar to the comparison article is to use the search engine of the search comparison mechanism, and compare The component and the determining component compare the minimum detected sentences filtered by the disassembled sentence algorithm one by one, and the search results obtained after the search are compared according to the searched sentence and the search result. To summarize, calculate the proportion of the common subsequences, and set a threshold. If the proportion of the common subsequence exceeds the threshold, the plagiar source comparison component is linked and the webpage or article is downloaded. Solve some of the same pages or articles in the article for subsequent follow-up comparison procedures. According to the Chinese digital anti-plagiarism detection and comparison system described in the first or second patent application scope, the method of comparing the full text is to use the comparison component of the search comparison mechanism, and compare the article with the same webpage or The article is directly connected with meaningless characters, processed by the data structure of Suffix Array, and then using the technique of Data Partitioning (DP), which can get the longest part in the comparison article. There are sentences on the same page or article; after that, the judgment component of the search comparison mechanism is used to evaluate and judge the proportion of plagiarism after the full text comparison. According to the Chinese digital anti-plagiarism detection comparison system and method described in item 1 or 2 of the patent application scope, the comparison of the article and the identical network or article is the same, and the search comparison mechanism will be in an anti-yellow manner. Marked in comparison pages and similar pages or articles. According to the Chinese digital anti-plagiarism detection comparison system and method described in claim 1 or 2, wherein the central server is provided with a plurality of computing hosts, and the matching articles uploaded by each user are distributed to the central server. Each computing host uses the computing host to disassemble the article, download multiple pages or articles that are identical to the comparison article, and compare the full text to obtain the comparison result, and return the comparison result to the central server. Transfer to the user via the central server.