Efficient approach to detect logical equivalence in the paradigm of software plagiarism

Abstract

Software plagiarism is one of the major obstacles which complicates the development and deployment of software systems. Legal complications related to use of copyright protected code has a far reaching effect. The challenge is to detect the existence of logical equivalence in an otherwise seemingly unmatched code. In this paper, a novel approach has been described which detects plagiarism only in logical statements of the code i.e. it does not detect copy-pasting in inclusion statements (e.g. #include in C/C++, import in JAVA), macro definitions (e.g. #define in C/C++), declaration statements and inside commented code. It also takes into account possible cosmetic changes of the code e.g. variable renaming, data type alterations, code snippets insertions or minor modifications. The method does comparison in an efficient minimalistic way, thus making it scalable to large code bases.