Analysis And Research Of The Core Structure Of The Software Network Based On Inheritance Tree

With the development and growing maturity of object-oriented techniques and tools, compared with the structural design object-oriented design shows great advantages. Meanwhile, the traditional metrics of object-oriented software hardly reflect the basic characteristics of the system. Therefore, it needs new metrics to quantify the characteristics of these systems.Combined with the traditional object-oriented software metrics methods, and taking the open source software as objects, this paper proposes class hierarchy object-oriented software network model to study features of object-oriented software.Firstly, this paper designs and implements a metric analysis software network platform which is based on inheritance tree. The platform is able to analyze the software source code, extracts the corresponding software network and the relationship of inheritance to generate inheritance tree and at last calculates the value of the network topology.Based on analysis tools of the core structure of the software network and observing the software in the view of inheritance, the paper find that network nodes gather together by the units of inheritance tree, and this gathering only concentrates in a limited inheritance tree. By Statistics, this paper find that20%of nodes in the entire software network (class number)compose the core structure of the software. Through analyzing the distribution of the core inheritance trees (nodes and edges), Corresponding edges of the core inheritance tree have high coupling which is caused by the high cohesion of internal nodes of the inheritance tree. Using the method of Fetion and Melton to measure the coupling of the core inheritance tree, This paper also find the overall coupling measurement is high (usually more than5),which is the appearance of tight coupling. In addition, this paper has found there exists linear relationship between the degree of inheritance tree and the number of cores with the help of counting and analyzing of the distribution of the characteristic values of internal nodes of the core inheritance tree. It concludes there is close inter-relation between the software functionality and reusability. The power distribution of Clustering coefficient, degree and the relationship shows there is modular hierarchy structures within the inheritance trees.With the help of analyzing the core structure of software, it becomes available to grasp the overall framework of software and lessen the complexity of analyzing the software system, which have great influence on software metrics.