For any document that needs to be detected, the system first carries out hierarchical processing, and creates fingerprints according to the levels of chapters, paragraphs and sentences. At the same time, it also uses the same technology to create fingerprint indexes for the comparative documents in the resource library.
This hierarchical multi-level fingerprint structure can not only meet our rapid detection of long documents, but also meet the high requirements of the system for accuracy and recall because our minimum fingerprint granularity is sentences.
In principle, as long as the same sentence exists between the test document and the comparison document, and the semantic level reaches the system weight standard, it will be marked as red and included in the total text reproduction rate.
Extended data
brief introduction
Paper detection service, also known as paper duplicate checking, is a computer software detection system to deal with academic misconduct (including plagiarism, forgery, tampering, improper signature, multiple submissions, etc.). ) papers (including dissertations, academic papers, published papers, paper titles, scientific research achievements and student compositions).
China national knowledge infrastructure
HowNet system adopts adaptive multilevel fingerprint (AMLFP) feature detection technology independently developed by CNKI, which has the characteristics of fast detection speed, high accuracy, high recall rate and strong anti-interference. Support the detection of chapters, paragraphs and sentences at all levels;
Support document deformation detection such as document rewriting and multi-document combination; Support the detection of academic misconduct in long documents such as papers and books.
Baidu Encyclopedia-Paper Detection Service