Generally speaking, the repetition rate = the number of words plagiarized in the paper/the total number of words in the paper. Taking hownet as an example, hownet's paper detection includes several duplicate checking subsystems, but the calculation rules of these duplicate checking subsystems are the same, in other words, the calculation rules of hownet's paper detection rate are the same.

As far as the full text is concerned, when a paper is submitted for testing, HowNet will divide the content of your paper. For example, according to the sentence or several words as a region, this part is extracted and compared with the contents of the document library of the paper testing system. If there are similarities, they will be marked. Generally, 7-8 words are plagiarism. Of course, this is not absolute, and each system will be a little different. This explanation is the easiest to understand.

In terms of chapters, HowNet has the duplicate checking rate of each chapter in addition to the duplicate checking rate of full text and cited documents. The duplicate checking rate of each chapter refers to the number of repeated words in this chapter divided by the total number of words in this chapter, which shows the duplicate checking rate of each chapter.

At present, the detection rate of hownet papers can be found not only in the text, but also in the repetition rate of codes, formulas, tables and even foreign languages. Therefore, if HowNet checks the duplicate, this part is also included.