A New Coupling Metric: Blending Structural and Semantic Relations

Abstract

Coupling metrics represent the relationships between source code artifacts in software systems. Coupling is considered an important concept in measuring design quality and maintainability. Many coupling measures have been proposed in the context of object-oriented systems. A popular way to measure coupling is through structural properties and static code analysis. Another way to measure coupling is through semantic information encoded in identifiers and comments. However, there is still much to be understood about which aspects of coupling affect quality or other external attributes of software. This paper presents a new coupling metric for object-oriented systems that analyze both structural and semantic relationships between methods and classes. A study is performed on open source software systems to compare the new metric with existing structural and semantic coupling metrics. The study shows that the new metric captures new dimensions of coupling, which are not captured by existing coupling metrics. This paper investigates the use of the new proposed coupling metrics during change impact analysis. By comparing our new metric to other coupling metrics, we show that our new metric is a better predictor for classes impacted by changes.