A comparative study of static CIA techniques
Xiaobing Sun,Bin Li,Bixin Li,Wanzhi Wen
DOI: https://doi.org/10.1145/2430475.2430498
2012-01-01
Abstract:Software Change Impact Analysis (CIA) is an essential technique to identify the unpredicted and potential effects caused by software changes. A rich body of different CIA techniques, especially static CIA techniques, have continuously emerged in recent years. However, it is difficult for researchers or practitioners to decide which technique is most appropriate for their needs, or which CIA technique is more effective. Unfortunately, there was only a few work on the comparison of the CIA techniques. This paper presents a comparison study of different types of popular static CIA approaches, i.e., structural static analysis, textual analysis, and historical analysis. For each kind of static CIA approach, we introduce a representative technique, that is FCA -- CIA, ROSE, and IRC2M, respectively. Finally, some empirical studies are conducted on three real-world programs to compare the accuracy of these CIA techniques based on the precision and recall metrics. The results show that the accuracy of these three CIA techniques is different, and FCA - CIA has the best precision while the IRC2M has the best recall.