A Study of Test Case Prioritization for Regression Testing: The Case of Software Testing of Mobile Devices

• Main Authors: Chih-Chien Chang, 張志堅
• Other Authors: Eric T.G. Wang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/99489825301959507678

碩士 === 國立中央大學 === 資訊管理學系在職專班 === 101 === Software testing is a strenuous and expensive process. In order to reduce the cost of regression-testing activities, we can adopt test case prioritization techniques (TCP). Test case prioritization is an effective and practical technique applied in regression testing. In a survey of test case prioritization for regression testing, Qu et al. (2009) indicate the current techniques can be divided into code coverage-based (statement, branch or other coverage) and non-coverage-based. Some studies have considered the importance of characteristics of non-coverage-based for testing cases in test case run time and fulfillment of test case requirements. But for the problems such as how to consider the combined effects of various characteristics on testing cases, the existing research has not been able to provide a good answer. System testing for mobile devices is based on the system specification without inspecting the code, tested from the user’s point of view, which is also known as functional testing. The design and analysis of test cases is suitable for the non-coverage-based techniques. This study focuses on applying the non-coverage-based technology to comprehensively examine how important the characteristics of the test case affecting testing results. The results of the study should provide a model useful for software testing practice. In the study, we found the relationship between these important characteristics and management of company. We divided these characteristics into two types, one type belongs to the specification of company, and the other type belongs to enficiency of company. To further analysis, we found the characteristics belong to the type of enficiency of company that show fuzzy symptoms, we use these sympotoms to importve our model. Besides, we also found it is more appropriate to consider the fault detection capabilities of test cases with the time value of fault than with the ration of fault.