연구 분야: Software Development
학회: International Conference on Computing and Data Science
Test-to-code traceability links describe the test relationship between the test codes and the product codes. It is beneficial for maintaining the test codes during developing, but to manually create and manage such links are time-consuming and fragile. The automatic link establishing strategies can properly solve such an issue. Our works analyse the feasibility of using the co-evolution relationship in the code repositories to establish links between test and tested codes. Because the performance of such an idea depends on how the developers maintain their test codes, we mined six projects to observe how their developers did. The results demonstrated the rareness of such a relationship, which implies the bad performances of the idea. We still implemented the Co-Evolution and the three optimisations: co-creation relations, the commits filters, and the APRIORI algorithm. Then, we applied all these approaches to three large, well-studied open-source Java projects. The precision, recall and f1 score were used as performance measures during the experiments. Although all of them performed worse results than the previous approaches, the Co-Evolution approach shown a significant boost after these optimisations. That implies the potential of such kind of idea that to use the information from version control tools to establish the test-to-code traceability links.
| 발행 연도 | 2022년 |
|---|---|
| 인용수 | 0 |
| 출판 국가 | |
| 사이트 | Springer |
| 좋아요 수 | 0 |