Test cases reduction using similarity relation and conditional entropy

• Main Author: Md Nasir, Noor Fardzilawati
• Format: Thesis
• Language: English; English; English
• Published: 2017
• Subjects: QA76 Computer software
• Online Access: http://eprints.uthm.edu.my/839/

Testing is essential in software engineering due to the huge size and complexity of everyday software systems. Generation of effective test cases becomes a crucial task due to the increment in the source code size and rapid change of requirements. New test cases are generated and added to a test suite to exercise the latest modification to software. Therefore, it is not easy to select effective test cases due to their redundancy and having common requirements. Thus, new challenges arise in reducing redundant test cases and finding common requirements that would decrease the cost and maintenance of a software testing process. Given a test suite and a set of requirements covered by test suite, Test Case Reduction or Minimization aims to select a subset of test cases that covers the same requirements. Several techniques have been proposed by researchers based on reduction parameter such as Test Suite Reduction, Fault Capability Detection and Processing Time Reduction. Nonetheless, these techniques are unable to tackle all parameters simultaneously, for example, some techniques may perform well in reducing the size of test cases but less considered on fault detection ability. To address this issue, this study proposed a technique that is able to minimize the size of test cases and common requirement attributes without compromising on fault detection capability. The proposed technique uses Similarity Relation to reduce the size of the test cases and Conditional Entropy to reduce the number of common requirements. The experimental results show a test case reduction that is smaller in size without affecting the decision of the testing. The proposed technique was able to reduce up to 50% of the reduction rate compared to base-line techniques such as MFTS Algorithm, FLOWER, RZOLTAR and Weighted Greedy Algorithm.