概要: Software testing plays an essential role in software quality assurance. It helps developers to reveal and remove bugs in software. Developers often use test coverage to measure the sufficiency of tests, find non-tested statements, and localize a faulty statement. Traditional coverages, such as statement and branch coverage, are widely known and used. On the other hand, researchers have proposed various metrics for measuring test coverage of source code. Because such novel coverage metrics are not organized, it is impossible to understand and compare the benefits and limitations of each metric. This paper organizes the characteristics of each coverage metric by surveying a body of 43 papers that propose coverage metrics. The survey results showed that the novel metrics could be divided into two main groups: (1) metrics that improve or complement traditional coverage and (2) metrics that are effective in specific domains, such as concurrent programming. We performed a comparative analysis to identify the characteristics of each metric, such as benefits of use, effective domains, and information needed to measure coverage. Furthermore, we provide a catalog of coverage metrics to help developers and researchers select the best metrics for their context. タグ: code, coverage, metrics, preliminary, results, review, source, systematic

@inproceedings{谷口真幸2023,
  author = {谷口 真幸 and 柗本 真佑 and 楠本 真二},
  title = {A Systematic Review of Source Code Coverage Metrics: Preliminary Results},
  booktitle = {電子情報通信学会技術報告},
  volume = {122},
  number = {330},
  pages = {78--83},
  year = {2023},
  month = {jan},
  url = {https://ken.ieice.org/ken/paper/202301113CQj/}
}