Analysis and Comparison of Weighted Combinatorial Algorithms for Test Suite Reduction

• Main Authors: Chiu, Chang Yu, 邱昶羽
• Other Authors: Huang, Chin Yu
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/83683943756115333109

碩士 === 國立清華大學 === 資訊工程學系 === 104 === Using software has become a very important part of our daily life. Therefore, strict and rigorous development of software is necessary for developers. Software testing should be conducted carefully in the development process for minimal errors and ease of product usability. With the continuous functionality updating of software systems according to customized requirements, new test cases have been generated and included in the existing test pool. Finally, the size of test pool has often been too large, which costs large amounts of time to produce inefficient regression testing. Test suite reduction is one of the well-known issues, which is used to solve the size problem by removing redundant test cases. Following such, the test pool size can be reduced, while the remaining test cases are still able to provide the same coverage as the original test pool. However, most of the existing test suite reduction methods have considered only one or two testing criteria with no equivalence between them. In this paper, we propose three modified weighted combinatorial algorithms to flexibly and simultaneously consider two different types of testing criteria. Further, we also use a genetic algorithm to find the best weighting factor value assignment for each testing criterion. Experimental results show that our approach can keep nearly the same suite size reduction percentage, while significantly enhance the fault detection effectiveness in the selected representative test suite subset.