Empirical Exploration of the Relationship Between Coupling and Method Interaction-Based Cohesion Measures for Object-Oriented Classes

• Published in: 2023 Computer Applications & Technological Solutions (CATS) pp. 1 - 5
• Main Author: Al Dallal, Jehad
• Format: Conference Proceeding
• Language: English
• Published: IEEE 29.10.2023
• Subjects: class cohesion; class coupling; Correlation; Couplings; Guidelines; interaction-based cohesion; object-oriented code; quality measure; Software; Software engineering; Software measurement
• DOI: 10.1109/CATS58046.2023.10424366

Developing high-quality systems is an essential goal in software engineering. Coupling and cohesion are two highly considered quality attributes. For object-oriented systems, class cohesion refers to the relatedness of members of a class, whereas class coupling refers to the relationship between a class and other classes. To maintain a high-quality level, developers must aim at increasing class cohesion and reducing class coupling. Existing cohesion measures follow different measurement approaches. Interaction-based cohesion measures consider the degree of interaction between each pair of methods in a class. Aiming at providing quality-based guidelines for software developers, this study empirically explores the relationship between eight coupling and five interaction-based cohesion measures. The study involves classes in three open-source applications. The study results indicate that cohesion and coupling measures are inversely correlated. In addition, the results show that the correlation between cohesion and coupling measures is mostly weak, which empirically supports the expectation that these measures quantify different quality aspects. The strength of the correlation dramatically differs from one cohesion measure to another due to the differences between these measures in their formulas and measurement approaches.