Test Case and Requirement Selection Using Rough Set Theory and Conditional Entropy

The growing size and complexity of the software system makes testing essential in software engineering. In particular, the effectiveness of generating test cases becomes a crucial task, where there is an increment of source codes and a rapid change of the requirements. Therefore, the selection of effective test cases becomes problematic, when the test cases are redundant and having common requirements. Thus, new challenges arose to reduce the unnecessary test cases and find common requirements that would increase the cost and maintenance of the software testing process. To address this issue, this study proposed a technique that minimized the test cases and requirement attributes, without compromising on fault detection capability. The proposed technique, using Rough Set Theory-Similarity Relation, was used to reduce the size of the test cases. Subsequently, a new approach, known as Conditional Entropy-Based Similarity Measure, was introduced to obtain a minimum subset of requirements. It was anticipated that, the technique applied would contribute significantly towards solving the testing problems, since testers would no longer be required to select an arbitrary test suite on test runs. The proposed technique was found to have achieved up to 50% reduction of the processing time, as compared with the base-line techniques, such as, MFTS Algorithm, FLOWER, RZOLTAR and Weighted Greedy Algorithm.