The optimized grouping value for precise similarity comparison of dynamic birthmark

Abstract

Software birthmark is a technique for preventing and identifying plagiarism through illegal software modification using feature information from said software. Presently, research on birthmarks differentiates between static birthmarks and dynamic birthmarks according to the way of extracting feature information. Dynamic birthmarks are advantageous because such birthmarks can be extracted from a program without the source code. Because they are extracted from the executed program, however, dynamic birthmarks are susceptible to noise. Our previous research suggests that similarity comparison of dynamic birthmarks is an applied feature of dynamic birthmarks. In addition and through our previous research, similarity comparison of dynamic birthmarks were proven to have better performance than existing similarity comparison of birthmarks. Cosine similarities differed by grouping numbers, and the grouping number of the separated cosine was defined with a random number (1,000) in our previous research. In this paper, we find an optimized grouping value for similarity comparison of dynamic birthmark through experimentation. © 2014 ACM.