Graph Clustering Evaluation (GCE) metrics quantify the quality of clusters obtained by graph clustering (community detection) algorithms. In this paper, the authors argue that GCE metrics can be applied on graph representations of software systems in order to evaluate the degree of cohesiveness of software entities. In contrast to widely known cohesion measures used in software engineering, GCE metrics do not ignore external dependencies among software entities, but contrast them to internal dependencies to quantify cohesion. Using the theoretical framework of cohesion measurement in software engineering introduced by the researcher. They investigate the properties of GCE metrics.