Software structure characteristic measurement method based on weighted network

Abstract The function and quality of software is closely linked to its structure. One of the ways to study the structural features of software is the idea of applying complex networks. Software is abstracted into various network models for the purpose of study. Eigenvalues and additional parameters are then extracted from the model. Analyzing the characteristics and performance of the software by measuring the structural characteristics of the software and implementing a basic evaluation of the reliability and security of software. Due to the limitations of traditional measurement methods and most of the current measurement methods are software-based unprivileged networks, the unprivileged networks neglect the differences in the connection relationships in the software and their important attributes (coupling, aggregation). This paper will construct a directed weighted network model of software, and use this network model as a basis to redefine the calculation method of node connectivity, improve the calculation method of the node's degree of ripple, and then obtain the system's average degree of ripple as a measurement index. After experimental verification, the proposed method can more effectively quantify the structure of the software orderly and complex and make the evaluation of software features more accurate.