A hybrid AHP-DEA approach for software architecture evaluation and selection.

Saved in:

Bibliographic Details
Title:	A hybrid AHP-DEA approach for software architecture evaluation and selection.
Authors:	HAJIPOOR, Mehran¹, MOTAMENI, Homayun² h_motameni@yahoo.com, EBRAHIMNEJAD, Ali³
Source:	Turkish Journal of Electrical Engineering & Computer Sciences. 2025, Vol. 33 Issue 3, p224-247. 24p.
Subject Terms:	Analytic hierarchy process, Architectural style, Data envelopment analysis, Software architecture, *Software failures
Abstract:	Decisions made during the software architecture stage significantly influence the success or failure of software projects. Software architects must navigate the complex task of selecting the most suitable architecture style while balancing stakeholders' operational and nonoperational requirements. These requirements often involve diverse and conflicting quality attributes with varying priorities, as well as intricate interactions where some attributes positively or negatively influence others. Furthermore, different architectural styles exhibit varying levels of support for these quality attributes, adding another layer of complexity to the decision-making process. This study leverages the Analytic Hierarchy Process (AHP) and Data Envelopment Analysis (DEA) within a novel application framework tailored to software architecture evaluation and selection. The proposed framework addresses the complexities of evaluating trade-offs, interactions among quality attributes, and the varying levels of support provided by architectural styles. In the first phase, AHP evaluates the relative importance of quality attributes, their trade-offs, and their interactions with architectural styles. In the second phase, DEA employs optimization techniques using the AHP results to identify the most efficient architectural style. The results demonstrated that the proposed AHP-DEA framework effectively identifies optimal architecture styles by considering the trade-offs and interactions among quality attributes, achieving improved decision-making accuracy compared to traditional methods. The framework was validated in a real-world project, where it streamlined the evaluation process and provided reliable, actionable insights, ultimately confirming its practicality and effectiveness. [ABSTRACT FROM AUTHOR]
Database:	Energy & Power Source

Description
Abstract:	Decisions made during the software architecture stage significantly influence the success or failure of software projects. Software architects must navigate the complex task of selecting the most suitable architecture style while balancing stakeholders' operational and nonoperational requirements. These requirements often involve diverse and conflicting quality attributes with varying priorities, as well as intricate interactions where some attributes positively or negatively influence others. Furthermore, different architectural styles exhibit varying levels of support for these quality attributes, adding another layer of complexity to the decision-making process. This study leverages the Analytic Hierarchy Process (AHP) and Data Envelopment Analysis (DEA) within a novel application framework tailored to software architecture evaluation and selection. The proposed framework addresses the complexities of evaluating trade-offs, interactions among quality attributes, and the varying levels of support provided by architectural styles. In the first phase, AHP evaluates the relative importance of quality attributes, their trade-offs, and their interactions with architectural styles. In the second phase, DEA employs optimization techniques using the AHP results to identify the most efficient architectural style. The results demonstrated that the proposed AHP-DEA framework effectively identifies optimal architecture styles by considering the trade-offs and interactions among quality attributes, achieving improved decision-making accuracy compared to traditional methods. The framework was validated in a real-world project, where it streamlined the evaluation process and provided reliable, actionable insights, ultimately confirming its practicality and effectiveness. [ABSTRACT FROM AUTHOR]
ISSN:	13000632
DOI:	10.55730/1300-0632.4124