Author Identification of Software Source Code with Program Dependence Graphs

Abstract

With the significant increase of computer and Internet based crimes, it becomes increasingly important to have techniques that can be applied in a legal setting to assist the court in making judgements about malware, theft of code and computer fraud. To better deal with author identification of software, we propose a semantic approach to identifying authorship through the comparison of program data flows. To do so, we compute program dependences, compute program similarity if detecting theft of code is needed, and thus query about not only the syntactic structure of programs but also the data flow within in order to discriminate authors. The experimental result reveals that our technique is more robust even with some intentional code modifications.