Ordered Metric Methods of Classes Dependency Graph Based on Structure Entropy

Abstract

1 School of Information Engineering, Yancheng Teachers University, Yancheng 224002, China 2 Department of Medical Instrumen,Yancheng Third People\u2019s Hospital, Yancheng 224001, China (Received August 01 2019, accepted September 26 2019) In this paper, structure entropy which is based on classes dependency graph of a software system is \u00a0proposed \u00a0to measure \u00a0the \u00a0complexity \u00a0of \u00a0the \u00a0system. \u00a0In \u00a0order \u00a0to \u00a0research \u00a0the \u00a0structure entropy \u00a0metrics \u00a0of classes dependency graph in some \u00a0large \u00a0software systems, this paper takes the first \u00a0step to mathematically prove that structure entropy does not have the property of cohesion. Thus, a structure entropy can be used as a single \u00a0metric \u00a0of \u00a0complexity. \u00a0And \u00a0then, \u00a0a \u00a0program \u00a0is \u00a0written \u00a0based \u00a0on \u00a0relevant \u00a0matrix \u00a0algorithm \u00a0with \u00a0the construction of classes dependency graph. The corresponding metrics of structure entropy of three pieces of open source software are calculated and figured out based on classes dependency graph. The calculation shows that most classes in the \u00a0three pieces of the open source software \u00a0are \u00a0of randomness. Meanwhile, values of structure entropy features complex network Scale-free. Therefore, the different values of the structure entropy of open source software classes dependency graph influences the evaluation of software quality. Furthermore, some complex network statistical characteristics are found out in this paper, which will facilitate the further research on structure entropy as a metric of software complexity for sophisticated networks. \u00a0