Cloned Buggy Code Detection in Practice Using Normalized Compression Distance

Software developers often write similar source code fragments in a software product. Since such code fragments may include the same mistake, developers have to inspect code clones if they found a bug in their code. In this study, we developed a tool to detect clones of a faulty code fragment for a software company, since existing code clone detection tools do not fit the requirements of the company. The tool employs Normalized Compression Distance for source code comparison, because its definition is understandable for developers, and also it is easy to support multiple programming languages. We conducted two experiments using an existing research dataset and actual examples. Based on the evidence, the tool has been deployed in several projects in the company.