Methods And Realization Of Intelligent Visual Detection On Plagiarized C Source Code

Source code plagiarism detection is one important application of similarity code identification. Aiming at the demands of objectivity and authenticity about coding in the examination of computer course, this paper is to look for ways to copy the code C program identification method to achieve a target of visual detection with high efficiency and easy operation.By studying source code plagiarism types and identification methods, this paper puts two sides forward to identify and detection plagiarism within students'homework programs. First of all, because code of students has the typical features of simple, short length, single type, but in large, a space clustering method based on feature attributes vector is proposed. This method can extract representative attributes to build a vector space model, and then categorize plagiarism classes with the introduction of the vector space clustering algorithm, which is to identify suspects plagiarized source code in a relatively short period of time. Second on base of suspected plagiarized source files, in order to obtain a more intuitive visual result, from a regional perspective of plagiarized region recognition, this paper presents a measure of the SMR (Section Match Recognition) method based on structure matching technique. The method divides function section by structural analysis of the source file, and introduces RKR-GST string matching algorithm to match two source files on words level to identify sections with the similarity above threshold, which is to obtain a visual detection result of locating plagiarized region.On the basis of above research, a intelligent visual detection system on plagiarism C source code detection system is designed, together with code plagiarism detection evaluation system, which has provided an objective and reliable basis for evaluation. Tests show that the detection system can achieve significant effect on the common plagiarism types in assignments, especially ones of high similarity. The system also has been applied to the actual course examination and can identify 80% of the copied programs accurately, which has improved the quality of teaching and students learning, showing a strong practical value.