Graphs Resemblance based Software Birthmarks through Data Mining for Piracy Control

摘要

The emergence of software artifacts greatly emphasizes the need for protecting intellectual property rights (IPR) hampered by software piracy requiring effective measures for software piracy control. Software birthmarking targets to counter ownership theft of software by identifying similarity of their origins. A novice birthmarking approach has been proposed in this paper that is based on hybrid of text-mining and graph-mining techniques. The code elements of a program and their relations with other elements have been identified through their properties (i.e., code constructs) and transformed into Graph Manipulation Language (GML). The software birthmarks generated by exploiting the graph theoretic properties (through clustering coefficient) are used for the classifications of similarity or dissimilarity of two programs. The proposed technique has been evaluated over metrics of credibility, resilience, method theft, modified code detection and self-copy detection for programs asserting the effectiveness of proposed approach against software ownership theft. The comparative analysis of proposed approach with contemporary ones shows better results for having properties and relations of program nodes and for employing dynamic techniques of graph mining without adding any overhead (such as increased program size and processing cost).