Software birthmark based theft detection of JavaScript programs using agglomerative clustering and Frequent Subgraph Mining

Abstract

Use of JavaScript in web development has increased the theft of JavaScript programs. Mostly, JavaScript program are susceptible to theft because browsers provide the simplest way to access it. Techniques to safeguard software are code obfuscation and watermarking. A potential attacker can easily deface watermark and hence watermark cannot completely protect the code. Code obfuscation cannot avoid copy of code. It only prevents others by understanding the logic of the program. Our aim is to secure the intellectual property rights of JavaScript developers. Even if the code is obfuscated or a watermark is added in it, this system can easily detect theft. Heap Graph is used to depict the behavior of a JavaScript program as how it calls other objects to fulfil the desired functionality. Agglomerative clustering is used to efficiently merge the heaps formed during the phase of programs execution. Frequent Subgraph Mining is used for finding frequent set of nodes which represents the unique behavior of the program. Finally the subgraph of plaintiff program is explored contrary to graph of the suspected one. We worked on 3000 combinations of websites and found that the software is capable of finding even a minor theft.