Test case prioritization for object-oriented software: An adaptive random sequence approach based on clustering

• Published in: The Journal of systems and software Vol. 135; pp. 107 - 125
• Main Authors: Chen, Jinfu, Zhu, Lili, Chen, Tsong Yueh, Towey, Dave, Kuo, Fei-Ching, Huang, Rubing, Guo, Yuchi
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.01.2018
• Subjects: Adaptive random sequence; Cluster analysis; Object-oriented software; Test cases prioritization; Test cases selection; Cluster analysis; Test cases selection; Adaptive random sequence; Object-oriented software; Test cases prioritization
• ISSN: 0164-1212; 1873-1228
• DOI: 10.1016/j.jss.2017.09.031

•A new dissimilarity metric (OMV, object method vector) for OOS test cases.•Two clustering methods to generate adaptive random sequences for OOS test cases.•A sampling strategy to construct the adaptive random sequences.•Conducting empirical studies to verify the proposed approaches. Test case prioritization (TCP) attempts to improve fault detection effectiveness by scheduling the important test cases to be executed earlier, where the importance is determined by some criteria or strategies. Adaptive random sequences (ARSs) can be used to improve the effectiveness of TCP based on white-box information (such as code coverage information) or black-box information (such as test input information). To improve the testing effectiveness for object-oriented software in regression testing, in this paper, we present an ARS approach based on clustering techniques using black-box information. We use two clustering methods: (1) clustering test cases according to the number of objects and methods, using the K-means and K-medoids clustering algorithms; and (2) clustered based on an object and method invocation sequence similarity metric using the K-medoids clustering algorithm. Our approach can construct ARSs that attempt to make their neighboring test cases as diverse as possible. Experimental studies were also conducted to verify the proposed approach, with the results showing both enhanced probability of earlier fault detection, and higher effectiveness than random prioritization and method coverage TCP technique.