The Study And Realization Of Paper Plagiarism Identification System Based On The Text Structure

Plagiarism is becoming a more and more serious problem nowadays, and abundant resources in the form of number make plagiarism easier with the popularization and rapid development of digital library and the internet. Especially for students and academic researchers, they can easily find the relative information about the research topic through the searching tool on the internet. In particular recent years, plagiarism like copy and multiple submissions are quite common, and people attach great importance to the seriousness of the problem. In order to put an end to such phenomenon and purify the academic world, it is of great urgency to build effective copy detection system except enhancing educating the students and building relevant laws.To sum up the current copy detection technique and system, it is found that the defects are as below: first, the existing prototype system does not analyze the present common plagiarism form that one paper copies multiple papers. Second, the current prototype system does not compute the text similarity during the process of detection. Third, the existing prototype system does not give the relevant type of plagiarism discrimination to the text that already being regarded as plagiarism, and can not rapidly and accurately capture the obvious plagiarism type. In this context, the dissertation studies the existing copy detection technique, analyzes the features of plagiarism, and adopts finger printing method (similar to COPS) to identify the complete plagiarism and part plagiarism of the paper, with the method of word frequency which is based on the similarity of the text to identify the implicit plagiarism, and comparing the method before and after using P-R and MAP.