A Source Code Similarity System for Plagiarism Detection
A Source Code Similarity System for Plagiarism Detection
复制标题
用于抄袭检测的源代码相似性系统
DOI：
10.1093/comjnl/bxs018
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
D. Gašević
中科院分区：
文献类型：
--
作者：
Z. Duric;D. Gašević
源代码剽窃是一个容易做的任务，但很难检测没有适当的工具支持。已经开发了各种源代码相似性检测系统来帮助检测源代码剽窃。这些系统需要识别大量的词法和结构源代码修改。例如，通过一些结构修改（例如修改控制结构，修改数据结构或重新设计源代码的结构），源代码可以以几乎看起来真实的方式改变。大多数现有的源代码相似性检测系统可以混淆时，这些结构的修改已被应用到原始源代码。为了被认为是有效的，源代码相似性检测系统必须解决这些问题。针对这些问题，我们设计并开发了源代码相似度检测系统。为了证明所提出的系统具有所需的有效性，我们执行了一个众所周知的一致性测试。与JPlag系统相比，该系统在检测源代码相似性时，各种词汇或结构的修改被应用到抄袭代码显示出良好的效果。作为对这些结果的确认，独立样本t检验显示，我们使用的测试集和我们在35 - 70%的实际可用截止阈值范围内进行的实验的F测量平均值之间存在统计学显著差异。
Source code plagiarism is an easy to do task, but very difficult to detect without proper tool support. Various source code similarity detection systems have been developed to help detect source code plagiarism. Those systems need to recognize a number of lexical and structural source code modifications. For example, by some structural modifications (e.g. modification of control structures, modification of data structures or structural redesign of source code) the source code can be changed in such a way that it almost looks genuine. Most of the existing source code similarity detection systems can be confused when these structural modifications have been applied to the original source code. To be considered effective, a source code similarity detection system must address these issues. To address them, we designed and developed the source code similarity system for plagiarism detection. To demonstrate that the proposed system has the desired effectiveness, we performed a well-known conformism test. The proposed system showed promising results as compared with the JPlag system in detecting source code similarity when various lexical or structural modifications are applied to plagiarized code. As a confirmation of these results, an independent samples t-test revealed that there was a statistically significant difference between average values of F-measures for the test sets that we used and for the experiments that we have done in the practically usable range of cut-off threshold values of 35–70%.