Research On Key Nodes Of Software Network Based On Static Analysis And Dynamic Tracking

From the network perspective,the current software systems are increasing at an amazing speed in terms of size and complexity.As a result,software quality and safety problems have always been the focus of researchers.Therefore,understanding and measuring the software network and mining the key nodes that may be neglected in the software network are of great significance for maintaining software security and effective management of software.In this paper,the complex network theory is applied to measure the software network,identifying its key nodes and ranking them.The main work is following.Firstly,in order to obtain the information of software structure in different states,modeling methods of static software structure and dynamic software structure are discussed respectively.In the static state of software,the static code analysis tool is used to obtain the structural relationship between software functions,and the structural relationship is mapped to an undirected weighted network model.In the process of software dynamic execution,the tracing technique is applied to obtain the dynamic calling relationship of software functions and map it to a directed weighted network model.Secondly,according to the node structure attribute of undirected weighted software network,key nodes mining algorithm based on node correlation is proposed.Minimum degree algorithm,realized the meticulous division of network nodes.According to the node level building structure vector,considering the correlation between nodes and their neighbors,the key node metrics based on node correlation can be obtained.Thirdly,according to the behavior characteristics of the directionally weighted network nodes,this paper proposes a key node mining algorithm based on node similarity.In this algorithm,the traditional random walk algorithm is improved to make the walking process more consistent with the behavior characteristics of directed weighted nodes.The similarity value of nodes is calculated according to the steady-state similarity matrix,and then the measurement value of node importance based on node similarity is obtained.Finally,based on real open-source software,the proposed algorithm is compared with other algorithms to verify the effectiveness of the proposed method.In addition,the distribution and change rules of key nodes under different versions are summarized by comprehensively considering the maintenance and upgrade of software.