The Software Metric Method Based On The Characteristics Of The Software Weighted Networks

With the sharp increment in software scale and the increasing complexity of application environments, software architecture in large-scale software have a significant impact on software quality, the problem of structural complexity can lead to the quality of the software is difficult to control, and widely considered software metric method of the software have not adapted to the scale of development, and that for the measurement of the complexity of the software architecture is not mature enough. Therefore, new software metrics in order to meet the requirement of software metric is so important.Through the use of complex networks theory, it provides favorable means for the measurement of the complexity of the software structure that we take the large software system as a complex network structure consists of the relationship between each unit and take networks topology to represent the structure of software systems to establish a network model. The existing software structure of complexity metrics based on the characteristics of complex networks mostly base on unweighted networks model, and research has demonstrated that weighted software can be more accurately express networks software structure features.This thesis combine with theoretical knowledge of the complex networks to build the software weighted networks model, and give different weights to edge according to the relationship among nodes. Then we select some of the weighted networks software parameters for analysis and study the regularity, and establish the theoretical basis for the follow-up of the software structure metrics. And then we take hundreds of different sizes of object-oriented open source software to statistical analysis of the weighted networks, and get the relationship between the characteristics of the software networks topology and software architecture complexity through the statistical distribution of the characteristic value of the data. And then, based on the analysis of characteristics of the preliminary results we propose software structure metrics based on software weighted networks to measure the complexity of the software structure. We also compare with the traditional software structure measurement method. When measure the structure complexity of software weighted networks, we have a system evaluation for the software structure measurement method based on node strength and closeness combined with examples, and verify the validity of the proposed software structure complexity metrics. Finally, we use the weighted average of software networks shortest path length and the cost of the spread measure the transmission efficiency of software structure, and assess the risks of the software structure due to the modification changes and structural design of the module, thus improve the maintainability of the software quality. Combined with the example analysis, we test and verify the cost of the software structure based on transmission dissemination efficiency measure and risk assessment method is effective.