Alternative Process Options Software Testing
Keywords
Abstract
It is taken into account that the solution to the problem of software quality provision depends significantly on one of the final stages related to the level of testing of software products, which necessitates the development of alternative approaches to the testing process that would effectively take into account various factors influencing its effectiveness, the complexity of production and organizational conditions, and the requirements regarding the quality of the final product. An informational interpretation of the concept of quality formation of this category is proposed using the tools of decision theory to model alternative and determine the optimal variants of the program testing process. The application of an alternative approach to the selection of the software testing process options based on the method of linear convolution of criteria, the Pareto concept, taking into account the weight preferences of factors influencing the quality of software testing and their share in the formation of alternatives is considered. The theoretical fundamentals of the method of linear convolution of criteria and its application to program testing are presented. Practical implementation of the method is investigated, the initial stage of which is the selection of a Pareto set from the general set of characteristics of software testing, which contains the dominant factors of priority impact on the testing process. The use of the Pareto principle ensured the determination of the optimal set of combinations of test cases that best meet the needs of testing with limited resources. Alternative variants of the software testing process are designed using utility functions and efficiency shares of factors in alternative variants. The matrices of pairwise comparisons are constructed and processed to calculate the refined weight priorities of the factors of the Pareto set and the values of the utility functions of the factors, taking into account their share in the formation of alternatives. The optimal variant of the software testing process among the alternatives is calculated and determined according to the criterion of the maximum value of the combined target functions, which ensures the proper quality of simulation modeling.
