The Research Of Test-Suite Reduction Based On K-Medoids

Software testing is a significant activity to the software development process. Soft industry standardization, and constantly improve the technology also will drive the continuous development of the software testing industry. Software testing gradually evolved from the original pure manual test standardization procedures is also required of the work. As more and more profound understanding of the importance of software testing, the proportion of software testing phase in the entire software development cycle is increasing. During the software development process, because of the iteration, evolution, it is necessary to test the software frequently so that the test suite is becoming larger and larger. The cost of Test cases’design, implementation, management and maintenance is quite large, while the test resources are often limited; the reduction of the test suite has become a necessity. Ensuring the software is fully tested, how to generate a minimized test suite which can detect faults in the software as many as possible. To raise testing efficiency and reduce the cost of software testing is the key point in this paper.Test suite reduction problem aims to satisfy all testing requirements with the minimum number of test cases, so that the testing efficiency can be improved while the testing cost can be decreased. To address the issue, researchers have proposed a variety of reduction methods, such as the Heuristic Method (G Algorithm, HGS Algorithm, GE&GRE Algorithm, GA Algorithm), Integer Linear Programming, Demand-driven Approach, Genetic Algorithm, and so on. However, these methods are inadequate for some shortcomings:these methods have been shown that when the reduction reaches a certain degree, the fault detection effectiveness of the test suite will be greatly reduced, thus affecting the efficiency of software testing.Based on the analysis and summary of the algorithms which based on the basis of the test requirements set, this paper presented a test suite reduction algorithm which can get the test suite with less test cases and higher fault detection effectiveness. The algorithm uses one of the cluster analysis algorithms----K-medoids Algorithm. First, by clustering analysis of the algorithm on the original test suite, then, according to the testing requirements set, select the test cases in each cluster obtained, then, get the representative set which is the resulting suite called in this paper.Finally, according to the test suite reduction, use of the K-medoids algorithm to the realization of the code of experimental results show that the K-medoids algorithm based on the Cluster Analysis can more quickly get the small test suite. And the fault detection effectiveness is higher, After comparing the others classical algorithms, the experimental results show that the algorithm provided in this paper combining genetic algorithm with K-medoids algorithm could get the test suite quickly which has less test cases and the higher fault detection effectiveness.The experimental data proves that the algorithm is effective and feasible and it could be effective in improving the efficiency of software testing.