A Token Oriented Measurement Method of Source Code Similarity

In order to help teachers to identify plagiarism in student assignment submissions among students’ Source code quickly and accurately, this paper discusses a measurement method of Source code similarity. In the proposed algorithm, firstly, both of token oriented edit distance (TD) and token oriented length of longest common subsequence (TLCSLen) is calculated; secondly, considering the TD and TLCSLen, a similarity calculation formula is given to measure similarity of Source code; Thirdly, a dynamic and variable similarity threshold is set to determine whether there is plagiarism between Source codes, which ensure a relatively reasonable judgment of plagiarism. This method has been applied to the university's programming course work online submission system and online examination system. Practical application results show that this method can identify similar Source code timely, effectively and accurately.