SOFTWARE QUALITY FORMATION USING MULTICRITERIA OPTIMIZATION METHOD
DOI:
https://doi.org/10.31891/2219-9365-2024-77-14Keywords:
set of software quality factors, Pareto principle, alternative process execution variant, utility function of alternatives, intersection convolution of sets, convolution membership function, optimal implementation variant of the processAbstract
The quality of software is determined by various attributes and characteristics that affect its functional suitability, ease of use, performance or efficiency, maintainability, compatibility, reliability, security, and testing level. Ensuring high-quality software is a continuous process that requires planning, resources, and attention to detail at every stage of software development and maintenance. The effective implementation of this process determines user satisfaction and the success of the software product in the market.
In this article, the decision-making theory is employed to investigate and apply the information concept to address the issue of forming the quality of software. The standard only enumerates the conditions for compliance of software products with specified characteristics without disclosing the ways to achieve the result. The essence of the proposed approach lies in extracting from the general set of software quality characteristics a Pareto subset containing dominant factors that have a priority impact on the investigated process. To optimize the process of forming software quality during its design and development, a multi-criteria optimization method from operations research theory is utilized. This method uses fuzzy preference relations between alternatives to make decisions, the degree of which can be specified as a number on a unit interval.
Using the multi-criteria optimization method, fuzzy preference relations between alternatives are formed, identified by membership functions. Based on matrix analysis tools, utility functions and convolution membership functions are theoretically justified and calculated. These functions serve as a tool for establishing the optimal variant of software development process implementation. Using the convolution of relations and the set of non-dominated alternatives, the membership function of the final convolution is calculated, the maximum value of which determines the optimal alternative for the software development process.
