软件源码相似性分析技术及应用系统研究

发布时间：2018-10-25 11:34

【摘要】：计算机软件行业迅猛发展,有别于传统行业,它更可能出现知识产权纠纷,抄袭等。源代码作为计算机软件的核心价值体现,判定其是否有抄袭行为,这在司法取证等是一项非常必要的依据。检测源代码是否存在抄袭行为,也商业软件发布前判断是否使用了开源许可代码的重要依据,上线前的检测可以避免因违反某些开源代码使用许可而作为商业用途所带来的纠纷。代码相似性分析被广泛应用,可用于判断源代码是否存在抄袭,是否违反开源代码使用规则等。现有的工具往往使用一种比对算法,针对单一的语言,比对能力极为有限。本课题设计并实现了一套相似性综合分析系统,它使用了基于字符,标签序列和语法的三种核心比对,另外用整文件比对和部分语义比对作为辅助,并根据每种算法所占的比重设计出综合比对算法,比对极为精准。代码相似性检测是一种静态检测方法,主要用于检测源代码是否存在抄袭。本系统检测可以精确到行,并有多种先进成熟算法相结合,考虑到检测过程是使用大量样本或以开源代码为主的样本库,可以构建一个以缺陷代码为主的样本集,适用于检测其它软件代码的安全性。本文构造了大量的测试样本,涉及多种缺陷类型,对造成每种缺陷类型的可能性都进行了覆盖,通过对大量源码工程的比对,证明了该方法的优越性和可靠性。
[Abstract]:The rapid development of computer software industry, different from the traditional industry, it is more likely to appear intellectual property disputes, plagiarism and so on. As the core value of computer software, the source code is a very necessary basis for judging whether there is plagiarism or not. The important basis for detecting plagiarism in source code and determining whether open source license code is used before commercial software is released, Pre-line testing can avoid disputes over commercial use of certain open source code licenses. Code similarity analysis is widely used to determine whether plagiarism exists in source code and whether it violates the rules of open source code. The existing tools often use a matching algorithm, for a single language, the ability of comparison is extremely limited. This paper designs and implements a set of similarity comprehensive analysis system, which uses three kinds of core alignment based on character, tag sequence and syntax, and uses integer file alignment and some semantic alignment as auxiliary. A comprehensive comparison algorithm is designed according to the proportion of each algorithm, and the comparison is extremely accurate. Code similarity detection is a static detection method, mainly used to detect whether plagiarism exists in source code. The system can be accurate to the line, and has a variety of advanced and mature algorithms, considering that the detection process is to use a large number of samples or open source code based on the sample library, can build a defect code based on the sample set, Suitable for detecting the security of other software code. In this paper, a large number of test samples are constructed, involving a variety of defect types, and the possibility of causing each defect type is covered. The superiority and reliability of this method are proved by comparing a large number of source code projects.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP311.53

【参考文献】