The principle of CNKI duplication checking mainly includes two parts: text comparison and chapter-by-chapter detection.
Text comparison: HowNet duplication checking system will compare the text in the paper with the documents in the database to detect the similarity. In this process, the system identifies the text content in the paper and then compares it with the literature in the database. If the similarity is found to exceed a certain threshold, then this part of the content will be judged as duplicate.
Chapter-by-chapter detection: CNKI’s duplication checking system will also perform chapter-by-chapter detection based on the chapters of the paper. The content of each chapter of the paper will be compared separately, and the repetition rate of each chapter will be calculated. Finally, the repetition rates of these chapters are weighted and averaged to arrive at the final repetition rate for the entire paper.
In addition, CNKI’s duplication checking system will also take into account the structure and semantics of sentences and paragraphs when detecting text similarity. If a sentence contains a large number of quotations from other documents, it will be judged as a duplication even if appropriate citation annotation is made.
It should be noted that the CNKI duplication checking system is not completely accurate when detecting text similarity. For example, it cannot recognize non-text content such as pictures, tables, formulas, etc., nor can it recognize content in references. In addition, due to differences in citation methods and language expressions in different documents, misjudgments sometimes occur. Therefore, when using the CNKI duplication checking system, you need to make judgments and processes based on the content of your own paper and the actual situation.