A Novel technique for test case minimization in object oriented testing

Abstract

Software maintenance is the costliest phase of software life cycle and it consumes almost 70 percent of resources of development process. Software testing involves the examining the built software with the intension to find the defects in it. Exhaustive testing of the software for all possible test cases is not feasible as may take infinitely large amount of time and may consume large number of other resources. Researchers in the field of the software testing are exploring the different possibilities to reduce the required number of test cases to test given software. In case of an object oriented (OO) software, the complexities like inheritance, coupling and cohesion, make the software modules more prone to the faults. This problem gets augmented in case of very large software systems. Many researchers have solved the problem of test case minimization from the different perspectives like requirements coverage; statement coverage and risk coverage. But negligible research has been done of the basis of security metrics coverage. In this paper, a technique has been presented for minimization of test cases for the OO systems. The research reported in the paper has considered security as a perspective for test cases evaluation and minimization. Publically available data sets pertaining to open source software Camel 1.6.1 have been used for the evaluation of proposed methodology. Linear Regression (LR) model for the bugs present in the data and various object oriented metrics for security have been developed. The proposed model dissects the given metrics sets into effective and non-effective metrics. Effective metrics are then utilized for giving weights to the test cases, further with the help of weights obtained the test suite is minimized. The performance results of proposed approach are encouraging.