Detailed description of hownet system calculation standard:

1. After reading the introduction of this system, I have a question. This system is good for text copy recognition, but what about other contents, such as data and charts? Isn't it useless to detect it? Among all kinds of academic misconduct, plagiarism is the most common and serious. At present, the detection system has reached a high level. The detection of plagiarism and tampering of charts, formulas and data is currently under development and has made great progress. You are welcome to continue to pay attention to the progress of this detection system and put forward more critical and constructive opinions and suggestions.

2. According to this system, less than 39% is displayed in yellow, so does it mean that it is within the tolerable limit? Recently, I read the news that the national social science fund project of a teacher in Shanghai University was cancelled because two papers he published were plagiarized, accounting for 25% and 30% respectively. Please specify what the warning line is.

Percentage only describes the proportion of overlapping words in the detected documents, and does not refer to the plagiarism of the documents. It can only be said that the greater the percentage, the more overlapping words, and the greater the possibility of plagiarism. Whether it is plagiarism or not and the severity of plagiarism need to be decided by experts after review.

3. How to prevent the academic misconduct detection system of dissertations from becoming a platform for personal revenge?

This is something we are seriously considering. At present, this detection system is only used by users at the institutional level. We have established a strict management process. At the same time, technically, we have also taken various measures to prevent malicious acts as much as possible, including a series of strict identity authentication and login.

1. Title: accurate, concise, eye-catching and novel.

2. Table of Contents: A table of contents is a brief list of main paragraphs in a paper. (Essays don't need to be listed in the table of contents)

3. Abstract: It is an excerpt from the main content of the article, which requires short and pithy content. The number of words can be as few as dozens, and it is advisable not to exceed 300 words.

4. Keywords or subject words: keywords are selected from the title, abstract and text of the paper, and are words with substantive significance to express the central content of the paper. Keywords are words used by computer systems to index the content characteristics of papers, which are convenient for information systems to collect and provide readers with retrieval. Generally, 3-8 words are selected as keywords for each paper, and a new line is set at the bottom left of the "abstract".

Subject words are standard words. When determining the subject words, the paper should have a theme, and according to the indexing and collocation rules, it should be converted into standardized words in the subject glossary.

5. Text of the document:

(1) Introduction: Introduction, also known as preface, preface and introduction, is used at the beginning of the paper. The introduction should generally write the author's intention, explain the purpose and significance of the topic, and point out the scope of the paper writing. The introduction should be short and concise, and stick to the theme.

(2) Text of the paper: The text is the main body of the paper and should include arguments, arguments, argumentation process and conclusions. The main part includes the following contents:

A. Propose-demonstrate;

B. analyzing the problem-arguments and arguments;

C. solving problems-demonstrations and steps;

D. conclusion.

6. The references of the paper refer to the main documents listed at the end of the paper that can be referenced or quoted in writing. References should be marked on a new page according to GB77 14-87 "Description Rules of References at the End of Documents".

English: Title-Author-Publication Information (edition, publisher, publication date): Author-Title-The requirements for the references listed in the publication information are:

The references listed in (1) should be official publications for readers' textual research.

4. The minimum detection unit is one sentence, so you can't detect one or two words in each sentence?

We also deal with sentences accordingly, and have an algorithm of sentence similarity. It is not the same sentence that is judged to be the same. Sentences have sentence-level similarity algorithms, and paragraphs have paragraph-level similarity algorithms. Calculating whether a document or paragraph is similar to other documents is based on this.

5. If the original word is taken from relevant books, but the word has been copied from relevant documents in the database, that is to say, the previous article also took the same word from relevant books, but the words marked in my paper are from relevant books, is this academic plagiarism?

The detection system can't draw a conclusion, whether it is plagiarism or not, and finally there is manual review. So if it is the situation you describe, experts will make corresponding judgments. Our system only provides all kinds of clues and basis, so that people can quickly grasp the information of test literature.

6. The authority of HowNet detection system? The detection system of academic misconduct documents has not reached a conclusion, that is, the detection system does not characterize the detection documents, but only shows the similarities between the detection documents and other published documents and lists objective facts, and whether such detection documents belong to academic misconduct needs the final examination and confirmation of experts.