Research And Implementation Of Program Plagiarism Detection System

Nowadays, in higher education at home and abroad, homework plagiarism is widespread. Especially in computer courses, the program plagiarism is even more serious. Therefore, the research on program plagiarism detection is important.However, there still exist some deficiencies in the existing system and the existing detection algorithms. Most system only provides one detection algorithm, and the scope of application of detection algorithms is different, so users can not choose the right algorithm as needed. In actual testing, the paper also found a number of factors would bring the noise to the detection, impacting on the calculation of similarity. For example, a large number of duplicate variable declaration statement and the changes on the source code.This paper designed and implemented a source code plagiarism detection system, called PlagiarismChecker, which includes four detection algorithms. User can choose one or a combination of these algorithms and set a similarity threshold to filter the result. Also, users can easily add a new algorithm to PlagiarismChecker by implementing the interface. This paper provides a web application named Catcher, providing online program plagiarism detection services which result returned by e-mail. PlagiarismChecker is proved to be effective by experiments, because it can detect more plagiarism than the existing system.For the actual detection of the noise problem exists in the abstract syntax tree-based detection algorithm, the paper proposes using the normalized operator to solve this problem. Five normalized operator were designed and implemented and it proved that noise can be eliminated effectively in this way. This paper also improved PlagiarismChecker system, making the system can easily add a new operator, so that PlagiarismChecker can deal with the new discovery of plagiarism practices by adding a new operator.