Concern-Based Assessment of Architectural Stability: A Comparative Study

A concern is any consideration that can impact the software architecture and other artefacts. Sustaining stability of concerns realised in architectural designs is essential to obtain long-lasting software products. It is often assumed that concern assessment mechanisms are effective to predict architecture stability. They ought to exert a positive or detrimental effect on software stability. In fact, there is a growing variety of concern assessment mechanisms to support the quantification of concern properties in software artefacts. They range from metrics that compute basic concern characteristics – such as concern tangling and scattering – to patterns that represent recurring concern structures. However, the vast majority of concern assessment mechanisms has not been empirically validated yet. This paper presents a comparative study on the efficacy of a family of concern metrics and patterns to support architectural stability analysis. The correlation of concern properties and architecture instabilities is determined by quantifying the Spearman Correlation indicator. This study relies on several releases of two evolving software systems, where the architectural instabilities have been previously and independently documented.