An Empirical Study of Cohesion and Coupling: Balancing Optimization and Disruption

Abstract

Search based software engineering has been extensively applied to the problem of finding improved modular structures that maximise cohesion and minimise coupling. However, there has, hitherto, been no longitudinal study of developers' implementations, over a series of sequential releases. Moreover, results validating whether developers respect the fitness functions are scarce, and the potentially disruptive effect of search-based re-modularisation is usually overlooked. We present an empirical study of 233 sequential releases of 10 different systems; the largest empirical study reported in the literature so far, and the first longitudinal study. Our results provide evidence that developers do, indeed, respect the fitness functions used to optimise cohesion/coupling (they are statistically significantly better than arbitrary choices with p << 0.01), yet they also leave considerable room for further improvement (cohesion/coupling can be improved by 25% on average). However, we also report that optimising the structure is highly disruptive (on average more than 57% of the structure must change), while our results reveal that developers tend to avoid such disruption. Therefore, we introduce and evaluate a multiobjective evolutionary approach that minimises disruption while maximising cohesion/coupling improvement. This allows developers to balance reticence to disrupt existing modular structure, against their competing need to improve cohesion and coupling. The multiobjective approach is able to find modular structures that improve the cohesion of developers' implementations by 22.52%, while causing an acceptably low level of disruption (within that already tolerated by developers).

The Impact of Code Review on Architectural Changes

“…After filtering test code, we extract the MDG representing the structural architecture of the system for the versions before and after the revision. Previous studies that performed architectural analyses in Java systems relied on bytecode analysis for structural architecture extraction [20], [24], [38]- [42]. However, building and compiling the systems for each revision is a time consuming and error prone activity.…”

“…One might expect different results using different metrics. However, we rely on structural cohesion and coupling since they are widely-adopted for architecture analysis and have been thoroughly evaluated in previous studies [20], [24], [30]. Construct validity: We focus our analysis on the structural view of the software architecture, which is a low-level representation of the systems' architecture.…”

“…Recent studies have evaluated different metrics of structural cohesion and coupling as suitable measurements for architectural quality. In the context of search based software modularisation, Paixao et al [24] compared the modularisation developers implemented in their systems against baselines generated by different search procedures. The solutions implemented by developers outperformed most of the solutions generated by naive search procedures, indicating that structural measurements of cohesion and coupling are properties of interest for developers.…”

“…Quantitative studies evaluating metrics and techniques for structural optimisation [23], [24], [30] show how much architectural improvement can be achieved in software systems, but the feedback from developers is usually insufficient. Qualitative studies interview developers regarding architectural quality by either using toy systems [31] or selected past changes [19], [20].…”

Development and Optimization of NoSQL Database in Food Insecurity Early Warning System Based on Local Community Participation

“…In this phase, we check the quality of the logical data model by measuring its cohesion [22] [23]. Class cohesion refers to the relatedness of the class members and serves as one of the important aspects of the class design quality [25].…”