The stated contribution of this paper is to formally define metrics that help to analyze and evaluate coupling between classes. The following metrics are defined: coupling between entities (CBE), which includes CBE-in, CBE-out, and CBE-io (in-out); self coupling (SC); and total coupling (TC), which is the sum of these values.
The authors apply the metrics to the semantic Web technology evaluation ontology (SWETO) version 1.4. SWETO has 114 classes and a set of properties that includes 13 object properties and 56 data type properties. In this example, the authors use only the object properties. The object properties use 13 classes in the domain and range of the properties. According to the definition, a minimum of two classes or a maximum of 26 classes can define the object properties. The authors do not demonstrate the CBE-io metric in this example since SWETO has no inverse functions. They state that SWETO uses only 13 out of 114 classes, and this is considered low coupling. They further conclude: “We consider that high coupled ontologies would be desirable because low coupling would imply that classes are not related [to] each other except for ISA relationships.” They do not define the threshold for achieving high coupling.
The authors do not formally validate the metrics, nor do they evaluate their usefulness in clearly representing the ontology’s subject matter. Furthermore, they do not clarify the meaning of “important” in their statement that “the most coupled class also represents the most important one.” The domain values of the data type properties provide a very different nuance to the subject matter of the ontology, which the authors fail to capture in these metrics.