Software Topology Analysis And Complexity Measurement Based On Complex Network

With the rapid development of computer technology,people's demand for software system changes gradually.In the meantime,the dependence on application software grows sharply,which leads to the acceleration of software version updating.Moreover,the complexity of the software network will increase with the expansion of the network scale.The high complexity of software network is one of the main reasons that makes the risk as well as the quality and cost of software development more and more difficult to be effectively controlled.The traditional methods of software engineering evaluation are limited to the understanding and measurement of the software system structure.And they also lack of the overall understanding of its structure.Therefore,how to effectively understand the structure of software system and accurately measure its complexity has become a hot spot of scientific research.Previous research has shown that the software network presents the characteristics of a complex network.Therefore,based on the theory of complex network,this paper analyzes the topological structure and characteristics of software network from the global perspective.Above all,the interaction between units in a software system is extracted and abstracted into a network topology consisting of nodes and connected edges.Besides,according to the four important features of a complex network,this paper analyzes four well designed and representative software networks.And then it is found that their static topology contains the characteristics of scale-free network and Small World Networks.Secondly,the identification of critical nodes in complex networks is also an important direction in the study of the structure characteristics of complex networks.Based on the structure hole theory,it uses the degree attribute of neighbor nodes and secondary neighbor nodes as well as the topological relationship between neighbor nodes,it proposes a new local centrality algorithm based on structure hole theory and double degree which can measure the importance of network nodes.Meanwhile,it uses four large open source software networks to verify the algorithm.The results show that the method based on the structure hole theory and the importance evaluation of double-degree nodes is more effective,and the calculation accuracy is obviously better than the Burt network constraint coefficient.Although the algorithm only considers the local properties of the network,it is simple and low in complexity.Eventually,it is suitable for the measurement of large-scale networks.Ultimately,the complexity of the software network and the network structure entropy are analyzed,and the former degree structure entropy based on degree distribution and information entropy is improved.And this paper comes up with a new index based on degree structure entropy and aggregation coefficient,which is able to measure the complexity about the whole structure of complex networks.After we selected four special networks to verify,it shows that this method solves the defect which the degree structure entropy can not measure the complexity of complex network based on the single structure characteristic.Once again we chose four real software networks to verify the method,the results still prove that this method is effective and feasible.And this method is more suitable for large-scale networks with the lower computational complexity.